Yttria stabilized zirconia ceramics are commodities used in various applications such as biomedical, mechanical, sensors, solid oxide fuel cells, and thermal barrier coatings [1
]. For structural applications zirconia materials stabilized in the tetragonal phase by addition of 2–3 mol % Y2
(2-3Y-TZP) are relevant. These materials show a combination of high strength and fracture resistance due to transformation toughening, a stress induced martensitic transformation from metastable tetragonal to stable monoclinic phase. The phase transition comes along with a volume expansion ε = 4–5% and shear of ≈16% [6
]. The volume expansion puts the crack under compressive stress and thereby reduces the stress intensity at the crack tip. Starting powders of 3Y-TZP are today typically made by coprecipitation of oxides or hydroxides from yttrium and zirconium salts (typically chlorides) and a subsequent washing (chloride removal) and heat treatment procedures. In coprecipitated Y-TZP the constituents are homogeneously mixed at atomic level [7
]. The phase diagram shows that the typical composition 3Y-TZP is in the t + c field at sintering temperature (1300–1500 °C). The composition is supersaturated in yttria. Sintered at low temperature the material is, however, fully tetragonal segregation of yttria to the grain boundaries is responsible for solute drag and fine grain microstructure [8
]. Formation of cubic phase is typically observed at high temperature and long dwell. Reducing the stabilizer content leads to an improved toughness. A reduction of yttrium content to 2 mol % Y2
moderately enhances the toughness from 4–5 MPa√m for 3Y-TZP to 5–7 MPa√m depending on measuring protocol [10
]. Binner has shown, however, that a significant toughness boost occurs at <2 mol % Y2
]. Understabilized (<2.5 mol % Y2
) fully tetragonal materials are, however, very sensitive to sintering conditions, overfiring leads to grain growth and decomposition of the tetragonal phase [13
Alternative stabilization methods are coating or high energy milling of zirconia and yttria in appropriate ratio. These “coated“ TZP materials sintered at low temperature show higher toughness at somewhat lower strength due to the existence of a stabilizer gradient [14
]. Another method to produce stabilized and co-stabilized zirconia is the hydrothermal synthesis [17
Detonation synthesis at first sight seems just a variation of the coprecipitation technology. The zirconia and yttria precursors are mixed with explosives and detonated [19
]. Stabilized zirconia is thereby formed within an extremely short time increment at high temperature and pressure. In an earlier study, 3Y-TZP made from detonation synthesis powder was tested with respect to mechanical properties and low temperature degradation resistance [21
]. The material exhibited slightly increased toughness and threshold toughness coupled with a two-stage ageing behavior resembling a ‘mixture’ of coprecipitated and coated TZP. Recently a 2Y-TZP material based on detonation synthesized powder was tested which showed both extremely high strength and high toughness and a surprisingly low tendency to low temperature degradation (LTD) for a material of low stabilizer content [22
]. This study aims at finding the clue to this uncommon combination of properties by carrying out a detailed mechanical microstructure and phase composition analysis. A critical assessment of the methods was carried out in order to appraise whether the standard methods and calculation models can be safely applied when analyzing very tough Y-TZP materials.
2. Materials and Methods
Nanoscale 2Y-TZP starting powder manufactured by detonation synthesis was supplied by Innovnano, Portugal. The powder was processed as-received without any further compounding procedures. Cylindrical samples of 45 mm diameter and 2.2 mm thickness were produced by hot pressing (FCT Anlagenbau, Sonneberg, Germany) in boron nitride clad graphite dies. The furnace chamber was evacuated to <1 mbar then the die was heated up at a pre-load of 3 kN (≈2 MPa) to 1150 °C at 50 K/min. Between 1150 °C and the final sintering temperature the heating rate was reduced to 15 K/min and the axial pressure was increased linearly to 50 MPa. Final temperatures were varied between 1250–1450 °C in 50 K increments. The dwell at final temperature and maximum pressure was 1 h. Then the heater was shut off, the load was reduced to zero within 3 min and the die was cooled in the press. Four disks per sintering temperature were produced to obtain a sufficient quantity of samples for analysis. The disks were first manually ground to remove the surrounding grit and subsequently glued on sample holders. The samples were machined by lapping with 15 µm diamond suspension and polishing with 15 µm, 6 µm, 3 µm, and 1 µm on an automatic machine (Struers Rotopol, Copenhagen, Denmark). Density by buoyancy method, Youngs’s modulus, and Poisson’s ratio (by resonance frequency method, IMCE, Genk, Belgium) were measured on entire disks. The disks were then cut into bending bars of 4 mm width on a diamond wheel (Struers Accutom, Denmark). The sides of the as-cut bending bars were lapped with 15 µm suspension to remove the cutting affected zone, the edges of the bending bars were carefully beveled with a blunt 20 µm diamond disk. Vickers hardness HV10 tests were carried out with the remaining pieces (Bareiss, Baiersbronn, Germany, five indents each).
Bending strength σ4pt
measurements were carried out using a four point setup with 20 mm outer and 10 mm inner span. The crosshead speed was 0.5 mm/min. 11 samples (cross section ≈2 × 3.8 mm²) per batch were tested. The fracture resistance was determined by three indentation based techniques. Direct crack length measurement (DCM) of HV10 was carried out and evaluated according to the Niihara Palmqvist crack model [23
]. Indentation strength in bending (ISB) tests were carried out on four bending bars per batch. The bars were notched with a HV10 indent in the middle of the tensile side with cracks parallel and perpendicular to the sides. The residual strength was measured immediately afterwards in the same four point setup with a crosshead speed of 2.5 mm/min. Toughness was calculated according to Chantikul [24
]. The third method was stable indentation crack growth in bending (SIGB) [25
]. Here two bars were notched with 4 HV10 indents each. Indents were placed at a distance of 2 mm and aligned along the middle axis of the tensile side. Here the samples were exposed to a multistage loading process. The initial loading was carried out with one-third of the residual strength in the ISB test, the load was then increased in 10% increments of the residual strength until fracture occurred. The length of the cracks was determined with the microscope of the hardness machine after each loading step. The resulting SIGB plot of Ψ∙σ∙√c versus P∙c−1.5
shows three relevant stress intensities (Ψ = crack geometry coefficient, σ = bending stress, c = crack length, P = indentation load = 98.1 N). For the crack geometry coefficient, which is Ψ = 1.27 for a halfpenny crack, a value of Ψ = 1 was assumed to account for the flat profile of the Palmqvist cracks [26
]. The ordinate intercept represents KIC
. The kink in the curve which occurs at the stress intensity Kapp,0
when the cracks start to grow represents the R-curve related toughness increment. The threshold toughness is the difference KIC
The phase composition of the zirconia material and its tetragonality was determined by XRD (X’Pert MPD, Malvern Panalytical, GB, CuKα1
, Bragg–Brentano setup). In the fingerprint range between 27–33° 2Θ the areas of the monoclinic 111 and -111 peaks and of the 101 tetragonal peak were integrated and the monoclinic content was calculated according to the calibration curve of Toraya [27
]. Integration was carried out using HighScore software (Malvern Panalytical, GB). Monoclinic contents of polished material Vm,p
and in fracture faces of ISB tests Vm,f
were measured and the transformability Vf
was calculated as the difference Vf
. In the 72–75° 2Θ range the tetragonality and cell volume were calculated based on the position of the tetragonal 400 and 004 peaks [28
]. Based on XRD data values for transformation zone size h were calculated according to Kosmac [29
]. Transformation toughness increments ΔKICT
were calculated according to McMeeking with data for transformation zone size h, transformability Vf
, Youngs’s modulus E and Poisson’s ratio ν [30
]. Transformation toughness values were calculated assuming different transformation efficiencies X (from X = 0.27 predominantly dilatoric and typical for Y-TZP to X = 0.48 for fully developed dilatation and shear contribution such as in Mg-PSZ).
The starting powder was investigated by XRD as received and in annealed state (600 °C/1 h) to identify the phase composition and estimate the particle size by Scherrer analysis [31
The microstructure of the materials was investigated by studying thermally etched samples (1150 °C/10 min hydrogen atmosphere), the grain size was determined by line intercept method using Mendelson’s correction factor of 1.58 [32
Moreover, the structure of the tensile surfaces of bars fractured in ISB and in regular four-point bending tests were inspected by optical microscopy using differential interference contrast (DIC) in order to identify transformation zones indicating transformation related fracture.
As shown in the previous chapter 2Y-TZP materials derived from detonation synthesis powder show a unique combination of strength and fracture resistance. Fracture resistance values are not as high as claimed by the powder producer (15 MPa√m). Still a combination of 1500 MPa strength and >10 MPa√m toughness is considerable and comes close or even exceeds the limits defined by Swain and Rose [40
] based on compiled data (which were moreover mainly three-point bending strength and indentation toughness data [41
Putting the results from mechanical testing phase analysis and microstructure together leads to significant discrepancies between the toughness measured and the toughness expected from phase composition data. As shown above the threshold toughness of ≈4.5 MPa√m determined by SIGB is only slightly higher than for other Y-TZP materials made from coprecipitated powders [42
]. Assuming only transformation toughness as a relevant process zone effect the high toughness can only be explained if an outstandingly high transformation efficiency is assumed. This is, however, unlikely and in contrast to existing literature [40
]. Y-TZP materials from stabilizer coated powder may have much higher threshold toughness [43
], indications for a stabilizer gradient which may be associated with the very off-equilibrium powder production process by detonation were not detected by XRD [15
]. 400 and 004 tetragonal peaks were single peaks appearing at identical 2Θ values no matter which sintering temperature was applied.
Another surprising correlation was the lack of a strict coupling of transformability and grain size. Figure 11
a–d show the evolution of grain size with sintering temperature, the evolution of transformability and toughness with grain size and the correlation between toughness and the relevant parameters Vf
∙√h in the McMeeking formula [30
The observed grain sizes with changing sintering temperature are quite identical to the values determined by Lange [13
], the material here completely retains the tetragonal phase up to grain sizes of 460 nm while Lange reports a monoclinic content exceeding 10 vol.% above a grain size of ≈230 nm. The 2Y-TZP in this study seems thus surprisingly stable. According to Swain, we would expect the maximum toughness in 2Y-TZP of ≈9 MPa√m at grain sizes above 1 µm and under formation of transformation zones with h = 8 µm [44
]. For such materials the critical stress is 1300 MPa. In the present case, the transformation zone sizes determined by XRD at comparable toughness levels are considerably smaller and the critical stress is significantly higher. Transformability shows a steep increase between 1250–1300 °C sintering temperature. Then a linear dependency follows at higher sintering temperatures. Toughness, however, does not follow grain size, but forms an intermediate maximum. The same is true for the correlation of toughness vs. Vf
∙√h which actually shows an adverse trend at higher grain sizes. Toughness values obtained by ISB tests at very high crosshead speed do show a direct correlation to transformability, at lower test speed a part of the toughening effect seems to be lost due to subcritical crack growth.
All these correlations raise some doubts about the correctness of the methodology chosen and/or if there is something special in the studied material that has yet to be discovered.
The images of the transformation zones observed close to the crack front on the tensile side of ISB and bending test bars show that the transformed region—at least at the tensile surface—is far larger than the observable maximum transformation zone size (≈10 µm). As Sergo has shown, these transformation bands are wedge shaped so that the thickness of the transformed region varies from the tensile surface into the bulk [45
]. The XRD measurement measures an average value, however, transformed regions going deeper than 10 µm will be figured in with a capped value of 10 µm.
ISB bars showed a much more controlled fracture behavior due to the large defect introduced by indentation from which the crack propagates (the straight crack plane makes XRD measurements very convenient). The regular bending test samples fractured starting from some random defect of sufficient size. As in the case of ISB these samples either fail at lower stress exhibiting linear stress–strain behavior or they show transformation dominated failure at very high stress (>1500 MPa) with a nonlinear stress–strain behavior. In the latter case the transformation should occur prior to fracture. A closer look at the edges of fractured bending test bars also shows some secondary cracks oriented in ≈30° to the primary crack plane. These side cracks also have transformation zones which contribute to toughness (however less efficient) but cannot be measured by XRD. As the crack energy seems to be dissipated in a much larger volume than in the transformation zone adjacent to the primary crack it is not necessary to assume an overly high transformation efficiency. However, it seems that for materials with such extended transformation areas no convenient analytic method to calculate the transformation toughness increment based on XRD data is available.
Another point is the discrepancy between the measured grain sizes and the measured peak width which—at first sight—implies a constant domain size. The observed grain sizes in the microstructure, however, do not support this interpretation. Another reason for the broad peaks could be a certain variation of stabilizer distribution in the grains. In an earlier investigation, some indications were identified which would support this interpretation [21
]. The powder producers mentioned that the powder is collected in the detonation reactor at different locations so that in fact a certain scattering in powder properties may be present which is not fully equilibrated during sintering. Basu found higher toughness in mixed 2Y-TZP (made from 3Y + 0Y) than in coprecipitated 2Y-TZP and attributed this to larger grain size, broader grain size distribution and broader variation of stabilizer content [11
]. The as-received starting powder is predominantly monoclinic and seems very poorly crystallized. In as-received state the tetragonal phase is probably not stabilized by yttria content but rather by fine grain size [45
]. Therefore, it seems that in the initial powder zirconia and yttria are not yet present as a solid solution. A very moderate annealing process at only 600 °C is capable of partially converting the monoclinic to tetragonal phase. Hence, it seems that the reaction conditions in the detonation process are sufficient to precipitate and mix yttria and zirconia very intimately—but probably not homogeneously. However, the conditions seem not to provide sufficient thermal energy to form a proper solid solution.
The fact that residual strength values determined at different testing speed are distinctly different can be explained. Static and cyclic fatigue tests by Chevalier [42
] clearly show that the measurable toughness of 3Y-TZP strongly depends on crack velocity. Faster crack growth comes along with suppression of subcritical crack growth induced by hydrolysis of bonds at the crack tip, measurements in air typically show three stages. It is very likely that the doubling of the testing speed shifted the position in the V–KI
The combination of high toughness and strength in 2Y-TZP manufactured from detonation synthesized powder was confirmed. The material exhibits a slightly enhanced threshold toughness of 4–4.5 MPa√m compared to ‘regular’ coprecipitated material (3.5 MPa√m). The critical transformation stress of the material amounts to approximately 1500 MPa, above this stress the material tends to transform as indicated by non-linearity of the stress–strain curves and formation of visible transformation zones on the tensile side of bending bars. The transformation related stress intensity increment is high (≈6 MPa√m) and cannot be sufficiently explained by assuming a transformed layer of height h as can be determined by XRD. Evidently, the material is very strong due to the fine grain size and very high stress is required to trigger the full-fledged transformation. Then the energy is dissipated over a larger volume which may explain the high toughness without taking the influence of shear into account.
The reason for the enhanced toughness of this special material compared to coprecipitated 2Y-TZP (which requires much higher sintering temperature and grain sizes than the detonation synthesized TZP to reach the same toughness level) is yet not completely understood. The powder is produced under extreme non-equilibrium conditions, it is very fine and predominantly monoclinic. Partial incorporation of yttria into the lattice resulting in formation of more tetragonal phase is, however, already triggered under relatively moderate annealing conditions. It can be assumed that there is a certain inhomogeneity of yttria distribution and/or crystalline disorder. This argument is supported by the broad XRD peaks indicating domain sizes which are in contrast to grain sizes determined by SEM. Crystalline disorder of smaller domains may also be the reason for the high stress required to trigger the transformation. Once the transformation is triggered, large transformation zones typically known from Ce-TZP are formed. Deeper insight into the diffusion and ordering processes during sintering could be delivered by detailed TEM studies of materials sintered at different temperatures.