Mechanical Properties of an Extremely Tough 1.5 mol% Yttria-Stabilized Zirconia Material

Abstract

Yttria-stabilized zirconia (Y-TZP) ceramics with a drastically reduced yttria content have been introduced by different manufacturers, aiming at improving the damage tolerance of ceramic components. In this study, an alumina-doped 1.5Y-TZP was axially pressed, pressureless sintered in air at 1250–1400 °C for 2 h and characterized with respect to mechanical properties, microstructure, and phase composition. The material exhibits a combination of a high strength of 1000 MPa and a high toughness of 8.5–10 MPa√m. The measured fracture toughness is, however, extremely dependent on the measurement protocol. Direct crack length measurements overestimate toughness due to trapping effects. The initially purely tetragonal material has a high transformability of >80%, the transformation behavior is predominantly dilational, and the measured R-curve-related toughness increments are in good agreement with the transformation toughness increments derived from XRD data.

1. Introduction

Ceramics are inherently brittle materials. In the absence of R-curve effects, their strength (σ) a priori depends on the fracture toughness (K_IC) and the defect size (a) (σ = K_IC/√(πa)). Improving the strength of ceramic components is, therefore, a matter of sophisticated processing (compounding, shaping, sintering and machining) to reduce the defect size and material science to develop materials with higher toughness. According to Evans, the total toughness is determined by a combination of the intrinsic toughness (K₀) and the sum of the toughness increments (K_i) introduced by different reinforcement effects (K_IC = K₀ + ΣK_i) [1].

In partially stabilized zirconia materials, the dominant reinforcement effect is transformation toughening, a stress-induced martensitic transformation of the metastable tetragonal phase to the stable monoclinic phase associated with volume expansion and shear [2]. The phase transformation is triggered at the tip of a crack under tensile stress. As the crack proceeds, the compressive stresses on both sides of the crack resulting from a transformation-induced volume expansion close the crack and thereby reduce the stress intensity at the crack tip [3].

In order to retain the metastable tetragonal phase after sintering, stabilizer oxides are added, which form solid solutions with the zirconia and stabilize the high temperature phase by expanding the lattice (oversize dopants) with the cation radius r_c > r (Zr⁴⁺) and/or by the introduction of oxygen vacancies (aliovalent cations, such as Y³⁺, or other trivalent rare earth cations) [4,5]. The second prerequisite for retaining the metastable tetragonal phase is a sufficiently fast cooling process after sintering.

Depending on the type and amount of the stabilizer oxide, partially stabilized zirconia materials (either entirely tetragonal, as in TZPs = tetragonal zirconia polycrystals, or as tetragonal precipitates in a cubic matrix, PSZ = partially stabilized zirconia) show different features concerning strength and toughness. Yttria-stabilized zirconia, which is frequently applied in mechanical engineering and the biomedical field, is typically stabilized with 3 mol-% Y₂O₃ (3Y-TZP) and has a high strength of >1000 MPa but a moderate toughness of ~5 MPa√m [6]. Ceria-stabilized zirconia (with ~12 mol-% CeO₂ = 12 Ce-TZP) has a high toughness value but only moderate strength [7]. The stabilizer cations also affect the transformation behavior of the TZP material. In Ce-TZP, the large tetravalent ceria cations expand the zirconia lattice (which leads to a minor lattice distortion) but do not introduce oxygen vacancies. Ce-TZP, therefore, tends to transform in a cooperative (autocatalytic) manner, as, due to the high symmetry of the lattice, the monoclinic lamella can easily trigger the transformation of adjacent grains [8]. The stress–strain curves of tough Ce-TZP materials show characteristics analogous to work hardening in metals. The autocatalytic or burst transformation is interrupted when the remaining tetragonal domains are locally separated by transformed material or a second phase, i.e., when the transformation is exhausted [9,10]. Ce-TZP materials are typically very tough, but autocatalytic transformation is not very efficient [11]. It is noteworthy that in Ce-TZP, there is a significant transformation before the crack tip and not only in the wake of the crack. Such materials show R-curve or transformation-dominated failure, i.e., the material transforms prior to crack growth.

In yttria-stabilized zirconia, the trivalent dopants introduce oxygen vacancies for charge neutrality, reduce the coordination number of zirconium cations, and thereby distort the zirconia lattice significantly [12]; additionally there is a moderate contribution by lattice expansion. The symmetry reduction leads to a more localized transformation behavior. Individual grains with the right size and orientation to the applied stress transform. The transformation is predominantly dilational; burst transformation is not observed. This leads to materials with a moderate toughness but a very steep R-curve [13]. The plateau toughness is reached after a few micrometers in crack length. This behavior makes Y-TZP materials attractive for small and mechanically highly loaded components.

According to McMeeking and Evans (Equation (1)) [14], the two important parameters which govern the transformation toughness (ΔK_IC^T) are the transformability (V_f) and the size of the transformation zone (h). All other parameters, such as υ = Poisson’s ratio, E = Young’s modulus, and ε_T = transformation strain, are material constants. The parameter X describes the transformation characteristics, where X = 0.22 for a purely dilational transformation and 0.48 for dilation and shear. Y-TZP has predominantly dilational transformation characteristics (X = 0.27) [13].

ΔK_IC^T = X/(1 − υ) ∙ε_T∙E∙V_f∙√h

(1)

The transformability of a tetragonal grain depends on its size, its stabilizer content, and the constraint of the surrounding matrix [15]. In order to increase the toughness of Y-TZP, either larger grain sizes are produced by increasing the firing temperature or the stabilizer content is reduced [16,17]. Thus, for a given stabilizer content, a critical grain size exists above which spontaneous transformation during cooling is observed. The lower the stabilizer content, the smaller the critical grain size [18]. Thermodynamically, a reduction in the stabilizer content has another implication, namely, the t/t + c phase boundary at typical sintering temperatures of ~1400 °C is located at 2.5 mol-% Y₂O₃ [19]. Materials made from coprecipitated Y-TZP powders with yttria contents above this limit are supersaturated with yttria. A complete segregation of the cubic phase, however, requires a high sintering temperature and prolonged sintering [20]. Materials with yttria contents lower than 2.5 mol-% are entirely tetragonal. Yttria segregation to grain boundaries is less pronounced than in 3Y-TZP. Hence, their grain growth is not retarded to the same extent. Moreover, such materials are potentially more prone to low-temperature degradation (LTD) due to their low yttria content and the moderate segregation of the stabilizer to the grain boundaries. Paul tested different ultrafine-grain Y-TZPs with respect to LTD, and they found a clear tendency to accelerated ageing with a reduced yttria content [21]. Low-yttria Y-TZP materials, therefore, typically require very-fine-grain starting powders with excellent sinterability to obtain fully dense and fine-grain sintered Y-TZP.

Y-TZP materials from powders with less than 2 mol-% yttria are well documented in the scientific literature. Toughness values achieved in these lab-scale studies were considerable [22,23]. However, until recently, few reproducible and processible ultrafine powders were available. The practical application of extremely understabilized Y-TZP was probably considered too dangerous with respect to spontaneous phase transformation and component failure. Today, fine-grain Y-TZP powders with yttria contents between 1.5 and 2 mol-% are available from different manufacturers in reproducible quality and are provided not only as plain powders but also as ready-to-press (RTP) granulate formulations. Concerning the Tosoh ZGAIA 1.5YHT RTP powder used in this study, an earlier publication by Matsui [24] gave information about the suitable sintering temperature range (≤1400 °C/1 h), indicated a homogeneous stabilizer distribution, and reported an extremely high toughness value (>20 MPa√m, measured by indentation) and a high bending strength of 1300 MPa (3 pt bending). With respect to the experimental database of strength–toughness correlations compiled by Swain [13], this extreme toughness combined with a high strength was questionable. In another study on the same material, Imariouane [25] confirmed a high strength and a lower but still very considerable toughness value of 8.5 MPa√m (SEVNB method), which is within the theoretically accepted limits. They showed transformation bands, indicating a transformation-induced failure yet no non-linearity in the stress–strain curves. The susceptibility towards LTD seems moderate. In another paper, the sensitivity to sintering conditions was highlighted, whereby slow cooling leads to LTD in the presence of ambient moisture and does not require the presence of hot water or steam [26]. A summary of the strength and toughness values of various 1.5–3Y-TZP materials using different measurement protocols is given in in

Table 1. Strength and toughness data for Y-TZP with 1.5–3 mol% Y₂O₃ stabilizer.

Composition Mol% Y2O3	Strength 1 [MPa]	Toughness 2 [MPa√m]	References
1.5	1088, 3PB	14.5, DCM	[23]
		4.7, SCF
1.5	2600, P3B	13.8, DCM	[27]
1.5	1438, P3B	4.8, SEVNB	[28]
1.5	1270, 3PB	22, DCM	[24]
1.5	995, 4PB	8.5, SEVNB	[25]
	1320, 3B3B
2	994 3PB	6.4, SEVNB	[29]
2	1560, 4PB	9.6, ISB	[30]
	2088, P3B
	1670, 3B3B
2	not determined	5.9, DCM	[31]
2 (3Y + 0Y)	1270, 3PB	10.3, DCM
1.5 (3Y + 0Y)	630, 3PB	14.5, DCM	[32]
1.75 (3Y + 0Y)	680, 3PB	10.3, DCM
2 (3Y + 0Y)	1050, 3PB	8, DCM
2	1250, BA	7.7, DCM	[33]
2.5	1500, 3PB	7.8 ISB	[34]
2.7	1050, 4PB	11, ISB	[35]
3	1250, 4PB	6.2, ISB
3	1192, 4PB	4.92, CNB	[8]

¹ Abbreviations for strength measurements: 3PB = 3 pt bending, 4PB = 4 pt bending, P3B = piston on three balls, 3B3B = three balls on three balls, BA = biaxial. ² Abbreviations for toughness measurements: DCM = direct crack length measurement, SCF = surface crack in flexure, SEVNB = single-edge V-notched beam, ISB = indentation strength in flexure, CNB = chevron-notched beam.

.

As shown in Table 1, several techniques have been applied to measure the fracture toughness in zirconia. Almost all of them suffer from specific deficits. The popular DCM (direct crack length measurement) technique, with various protocols for calculating toughness from the wing crack length of Vickers indents, has been heavily criticized [36]. However, there are few alternatives for small scale samples. Critical points are, on the one hand, the unclear criteria for crack arrest and—especially in the case of tough zirconia materials—the trapping of cracks in the transformed zones around the indent [37]. In addition, the crack geometry, especially for short cracks, is unclear and deviates from typical model geometries. The cracks are often very flat Palmqvist-type cracks. In the case of extremely tough materials, the crack pattern may also be irregular, with one or more cracks missing completely. Sample preparation is also an issue, as crack lengths in the same material can differ significantly due to the presence of machining-induced residual stresses.

The determination of crack lengths is highly subjective. SENB and SEVNB require sharp notches to deliver reliable results. The radius of the tip should not be more than one order of magnitude smaller than the grain size. “Blunt” notches lead to an overestimation of toughness [38]. In the case of 3Y-TZP, the chevron-notched beam (CNB) test, according to the recent round robin, leads to quite reproducible results, with a less than 10% scattering between different labs [39]. However, conventional notching for CNB or SEVNB is still somewhat inconvenient in comparison to the placement of starter cracks by indentation. Tests based on monitoring either the evolution of the crack length or the residual strength in indentation-notched samples in bending tests offer the opportunity to easily notch and extend the cracks to a higher length out of the transformed zone around the indent. The R-curve-related toughness is then better represented than in DCM tests.

Still, indentation strength in bending (ISB) [40] tests may lead to overestimated toughness values, if the transformation and hereby crack trapping are very pronounced. A deviation from the ideal halfpenny crack geometry should also be considered. Cook et al. developed a hybrid indentation-based test to determine fracture toughness from the fracture stress and crack length of surviving cracks in multiple indented samples [37]. As the surface crack created by indentation is not stress-free, this method compensates for the residual stress stored in the sample derived from the indentation process. Based on this methodology, the stable indentation crack growth in bending (SIGB) procedure was elaborated on by Steinbrech et al. [41,42]. SIGB is not based on a single maximum toughness value and allows for the determination of the resistance to sub-critical crack growth (a detailed description is given in Section 2). SIGB, combined with a measurement on an indented and annealed sample (stress-free), according to Anderson, allows for the construction of a complete R-curve in the short crack region [43]. This method was successfully applied to test Ce-TZP-alumina composites [44] and, in one group, to determine the toughness of 2.5Y-TZP-alumina composites with different fractions of alumina [34]. Still, for materials with extreme crack trapping, the same problems as in the ISB test occur. In good accordance with thermodynamic considerations, Yu and Shetty have shown, for Ce-TZP, that transformation toughening can be reduced or even switched off, if the samples are indented at a high temperature [10]. Hence, a hot indentation method was tested for Y-TZP to avoid crack trapping and induce indentation cracks which are long enough to perform meaningful SIGB tests. An additional test protocol similar to SIGB is the envelope method by Braun and Lawn, which uses different indentation loads and which was successfully tested for the determination of the T-curves of different aluminas of various grain sizes [45]. A SIGB test using Knoop indents of different loads was elaborated on by Lube and Fett [46]. They showed that the fracture toughness can be read from the intercept of two SIGB curves determined at different loads, which eliminates extrapolation to infinite crack lengths, which is unrealistic considering the typical size of ceramic components. As already mentioned, the geometry of the crack is important. Halfpenny cracks have a geometry factor of Ψ = 1.27 [37], and this geometry factor is a linear factor in toughness calculations. Surface cracks in tough materials created by indentation have a flatter crack profile and a lower geometry factor. If the profile is known, the geometry factor can be calculated, according to Newman-Raju [47]. Strobl et al. described the methodology and offer an interactive calculation tool [48,49]. Another point is that plastic deformation directly below the indent may lead to deviations from the ideal geometry, leading to two semicircles: for ultra-tough materials, the notion of machining out the top of the indent to retain a surface crack does not apply due to the small crack sizes.

The 1.5Y-TZP used for this study is a good sample material due to its high toughness and transformability. There are four previous studies which compare different indentation-based and conservative toughness-testing methods. We attempt to understand more about the transformation behavior and failure characteristics and clarify why direct crack length measurements lead to drastically overestimated toughness values. Finally, we try to elaborate on a reliable indentation-based protocol to measure fracture toughness for fine-grained and transformable TZPs.

2. Materials and Methods

The powder used in this study was a spray-granulated ready-to-press (RTP) powder provided by a manufacturer (ZGAIA 1.5HT, Tosoh, Tokio, Japan). According to the datasheet, this powder is stabilized with 1.5 mol-% Y₂O₃, and 0.25 vol% of alumina is added as a sintering aid. The loss-on-ignition amounts to 3.9 wt%. The specific surface area S_BET = 16 m²/g. The powder is the same as in [24,25]. The received granulate was de-bindered in air at 600 °C for 3 h in order to stabilize the granule morphology and to avoid sintering and grain growth. The granules were then studied by scanning electron microscopy (SEM) to obtain information on the granule size and morphology as well as the particle size of the powder.

Quadratic plates of 35 × 35 mm² in size and 2.5 mm in thickness were pressed using a manually operated uniaxial hydraulic press (Paul Weber, Remshalden, Germany). The stainless-steel die has rounded corners and double-sided punches. The applied load was 150 kN, corresponding to 125 MPa in pressure. A total of 9 g of the RTP powder was weighed for each sample, and 16 plates were manufactured with identical pressing parameters.

The plates were subsequently de-bindered in air (60 °C/h to 600 °C, 3 h dwell, free cooling). Four plates per sintering temperature were then sintered in a dental furnace at 1250 °C, 1300 °C, 1350 °C, and 1400 °C (MIHM-Vogt HT speed, Stutensee, Germany). Heating was performed with 2 °C/min to 1200 °C and then with 1 °C/min to the final sintering temperature, the dwell was 2 h, and then the samples were cooled with the maximum cooling speed possible.

The sintered plates were then manually beveled with a 40 µm diamond disk and glued on sample holders. Machining included automatic lapping with a 15 µm diamond suspension (Struers Rotopol, Copenhagen, Denmark) on both sides. One side was polished for 30 min each using 15 µm, 6 µm, 3 µm, and 1 µm diamond suspensions to obtain a mirror-like surface. After machining, the thickness of the plates was 2 ± 0.1 mm.

The plates were cut into bending bars that were 4 mm in width using a diamond wheel (Struers Accutom, Copenhagen, Denmark). The as-cut bars were lapped on the sides to remove the cutting grooves and beveled at the edges.

Mechanical testing included Vickers hardness measurements (five HV10 indents per sample, 98.1 N, Bareiss, Oberdischingen, Germany), and the Young’s modulus was assumed to be 210 GPa, which is the typical value for a fully dense Y-TZP. Bending strength was performed in a 4 pt setup with a 20 mm outer and a 10 mm inner span, and the crosshead speed was set to 0.5 mm/min (Zwick, Ulm, Germany, 10 samples). Up to three bars were tested in a 3 pt setup to eventually create transformation bands which allow for the determination of the transformation stress. The transformation stress (σ_T) can be calculated from the distance of the first transformation bands (d) divided by the span length (l) and the failure stress (σ_T = σ_F ∙ d/l) [50].

As data on the fracture toughness of this material are controversial in the literature, various toughness-measurement protocols were applied. Indentation toughness measurements with direct crack length measurements (K_DCM) were performed to reproduce Matsui’s measurements [24]. On the leftover pieces of the samples, as many indents as necessary were placed (typically ~20 per sample) to collect 20 valid wing crack length values (due to crack trapping in the transformation zone around the indents, most HV10 indents only produce 1–2 wing cracks, and fully valid crack patterns were rare). The indentation toughness (K_DCM) was then calculated using the formula of Niihara et al., assuming Palmqvist-type cracks [51].

Four bars were notched with four indents each on the polished tensile side for toughness measurements by indentation strength in bending (K_ISB). The cracks were placed at a distance of 2 mm, so that they all fit within the inner span of the abovementioned 4 pt setup. Cracks of the indents were oriented parallel and perpendicular to the sample sides. Four indents were placed to compensate for the poor yield of valid indents. The residual strength was tested immediately after notching in the same 4 pt setup at a fast crosshead speed of 2.5 mm/min to avoid subcritical crack-growth effects. The calculation of (K_ISB) was carried out according to Chantikul [40].

Moreover, this dummy indentation method (only one indent leads to a fracture) offers the opportunity to calculate the fracture resistance (K_LWN) from the extension of the cracks of the longest surviving cracks. This was carried out by the procedure described by Dransmann [42], assuming a crack geometry factor of ψ = 0.95 (this value was chosen after inspecting some fractured bars). Palmqvist cracks are flatter than halfpenny cracks, with a geometry factor of ψ = 1.27, for which the method was originally designed by Braun and Lawn [45]. The same crack-shape correction was also applied to the ISB test.

Finally, the stable indentation crack growth in bending (SIGB) test was carried out with HV10 indents placed at elevated temperatures. It is known that transformability is temperature-dependent. Tsai found a complete elimination of transformation in Ce-TZP at 400 °C [11]. The idea was to produce valid indentations with wing cracks exceeding the size of the transformation zone and thereby eliminate crack-trapping effects [37]. With respect to the thermal vulnerability of the 1.5Y-TZP material [26], preliminary tests were carried out to identify the correct temperature to reliably induce cracks without damaging the material too much. A heating plate (IKA-Werke, Staufen, Germany) was set to 200 °C, and the samples were pre-heated on a zirconia support plate for 15 min. Then, the sample was transferred to the indenter device together with the support plate (to prevent too-fast cooling), and the four HV10 indents were immediately placed, as described above. Temperatures >200 °C led to spontaneous transformation. The SIGB test was carried out with an initial bending stress of 400 MPa. Then, the stress was increased in 100 MPa increments until sample failure. The crack length was measured after each loading step with the microscope of the hardness machine. The crosshead speed was set to 5 mm/min to avoid subcritical crack growths in the same 4 pt test setup.

For an evaluation of the SIGB test results, it is assumed that the indenter causes a certain stress intensity which leads to crack opening. After lifting the indenter, a residual (negative!) stress intensity (K_res) is stored in the sample. The indentation cracks are exposed to a total stress intensity (K_tot), which combines K_res and the applied stress intensity (K_app) (Equation (2)).

K_tot = K_res + K_app = χ∙P∙c^−1.5 + ψ∙σ∙√c

(2)

where P is the indentation load; χ is the residual stress coefficient; ψ is a geometry factor (e.g., 1.27 for a halfpenny crack); σ is the bending stress; and c is the surface crack length. The onset of the crack growth by applying bending stress requires that the applied stress intensity be K_app > K_app,0 = −K_res. With an increasing applied stress intensity, the crack grows until the sample finally fails. For a quantitative evaluation, ψ∙σ∙√c is plotted versus P∙c^−1.5. In the resulting plot, ψ∙σ∙√c rises at a constant crack length until K_app > K_app,0. Then, at a higher stress intensity, the crack starts to grow, and the curve rises with a slope (χ). In the plot, the Y-axis intercept represents the fracture toughness (K_IC) (at an infinite crack length), while the kink in the curve represents K_app,0. K_app,0 is the R-curve-dependent part of the toughness, and K_I0 = K_IC–K_app,0 is the threshold toughness, also called the resistance to subcritical crack growth [42]. With respect to the flat crack profiles, a geometry factor of ψ = 0.95 was chosen for an evaluation of the plot. Hence, the toughness values with a crack-shape correction are approximately 25% lower compared to the values assuming semicircular cracks (ψ = 1.27).

The SIGB test also allows the calculation of R-curves [42]. In the case of very steep R-curves, the initial region at the onset of the crack growth is difficult to investigate by SEVNB tests. For the given material, Imariouane determined a plateau toughness after a crack extension of less than 100 µm [25]. The initial part of the R-curve can be measured according to Anderson [43]. Here, bending bars are indented at an elevated temperature as before and then annealed at 1150 °C for 5 min to revert the phase transformation and to eliminate the residual stress caused by the indentations. Such indentation-based R-curves are typically measured in two steps: for the initial few microns of crack growth, indented and annealed samples are tested (without residual stress, K_res = 0 and K_tot = ψσ√c), and for longer crack lengths, K_res → 0, the data from the SIGB tests can be used. Combination of both tests leads to the complete R-curve.

Based on the assumption that the compressive transformation-related stress is the main component of the indentation-induced residual stress (K_res) in the vicinity of the indent, the notching at an elevated temperature will trigger significantly less phase transformations; this should significantly reduce K_res and, thereby, the possibility of crack trapping.

Scanning electron microscopy (SEM) images of the microstructure were made from polished and thermally etched (1150 °C/5 min, air) surfaces (FEI Helios nanolab600, Eindhoven, the Netherlands, SE 3 kV acceleration voltage). The average grain size was determined by the linear intercept method using the correction factor of Mendelson [52].

The tensile sides of the fractured bars were inspected for transformation bands by optical microscopy equipped with a differential interference contrast (DIC).

The phase composition was studied by X-Ray diffraction (X’Pert MPD, Panalytical, Eindhoven, the Netherlands, CuKα1, Ge-monochromator, Bragg-Brentano setup, accelerator detector). The samples were studied as fired after sintering and in polished conditions. The fractured surfaces of the 4 pt bending bars were also inspected. The monoclinic and tetragonal fractions were evaluated by integrating the monoclinic -111 and 111 peaks and the tetragonal 101 peak in the 27–33° 2θ fingerprint range. The volumetric contents were calculated from the peak areas using the calibration curve of Toraya [53]. The transformation zone sizes (h) were calculated from XRD data, according to Kosmac [54]. Transformation toughness values were calculated according to McMeeking (Equation 1) using a transformation efficiency factor of X = 0.27 [14].

The density of the materials was determined by the buoyancy method (Kern&Sohn, Lörrach, Germany).

3. Results

3.1. Powder Characteristics

Figure 1a,b show an overview and detail the images of the granulate. The granules shown in Figure 1a have an average size of approximately 100 µm. The granules show some characteristic granulate defects, such as satellites, dimples, and broken granules. The broken granules and the granules with dimples indicate that the granules are actually hollow. Some granules are odd-shaped or contain satellites, i.e., smaller granules attached to larger ones during spray drying. Such granulate defects are difficult to avoid, if ultrafine powders are processed. The fine fractions consisting of granules of ~30 µm and smaller can be estimated at <10 vol%. The detailed image in Figure 1b of the granule surface shows that the primary particles have sizes of 50–150 nm. The specified BET surface of 16 m²/g gives rise to the assumption that the larger particles are probably polycrystalline.

3.2. Mechanical Characterization

Density measurements show a density increasing very moderately from 6.0740 g/cm³ at 1250 °C to 6.0832 g/cm³ at 1400 °C. The manufacturer’s datasheet claims a density of 6.09 g/cm³ for material sintered at 1350 °C for 2 h in air. The material can, therefore, be considered fully dense (with a relative density of 99.7–99.9% for all the samples), and differences between the materials are not related to insufficient densification. Figure 2 shows the values for the Vickers hardness (HV10) and the bending strength (σ_4pt). The hardness stays invariant at 1215 HV10 at sintering temperatures between 1250 and 1350 °C and then slightly declines to 1200 HV10 at 1400 °C. The bending strength slightly declines with an increasing sintering temperature from 1070 MPa to 1020 MPa. The low scattering of the bending strength at the sintering temperatures of 1250 °C and 1300 °C is noteworthy (only ±25–40 MPa). The larger scattering at 1350 °C is caused by a single sample, which exceeded the average strength by more than 200 MPa within a population of samples with ~1000 MPa in strength. Without this single outlier, the strength would be 1012 ± 32 MPa.

Figure 3 shows the stress–deformation curves of the strongest samples of the individual series of 4 pt bending tests. Only weak indications of non-linearity were observed in most of the stress–deformation curves. Typically, in the range of 750–850 MPa, a slight bending is observed. The slope of the curve does not decline continuously after the initiation of transformation, as we may expect. The slope remains constant, either up to failure, or, as in the rare case of a sample sintered at 1350 °C, a second bending event occurs, after which the slope continues to decline. The latter effect was, however, only observed for a single sample with an exceptionally high strength. A majority of the samples fail without a second bending event. Again, this outlier is not characteristic of the sample population, as all other samples sintered at 1350 °C fail in a similar manner as the samples sintered at 1400 °C.

In the case of the 3 pt bending tests (not shown in detail), there is either no observable non-linearity (samples sintered at 1350 °C and 1400 °C) or the non-linearity occurs shortly before failure. Some samples sintered at 1250 °C and 1300 °C show a weak non-linearity at ~1000 MPa. The difference between the 4 pt and 3 pt bending tests is probably due to the different volumes exposed to the constant stress. In the case of 4 pt bending, many transformation bands (see Figure 4) are initiated simultaneously, while in the case of 3 pt bending, initially, a single band is produced, which leads to a much lower non-linearity, which is observable only under high stress conditions.

This assumption is confirmed by the presence of transformation bands on the tensile sides of the bending bars, as shown in Figure 4. At low sintering temperatures (1250 °C and 1300 °C), the bands are not very pronounced and are rather blurry, and at high sintering temperatures (1350 °C and 1400 °C), the bands become more pronounced. The spacing between the bands ranges from 100–400 µm.

The transformation stresses measured from the location of the bands in the samples from the 3 pt bending tests are shown in Figure 5. The transformation stress decreases with an increasing sintering temperature, as expected. Note that the transformation stress in 4 pt bending is lower.

Figure 6 shows the results of the uncorrected DCM and ISB toughness measurements. The K_DCM values are extremely high (15–19 MPa√m) and confirm the findings of Matsui [24]. The toughness maximum is found at 1350 °C. With regard to the difficulty in obtaining valid indents, it is suspected that the trapping of cracks in the transformation zone surrounding the indent is responsible for overestimated toughness values [37]. A priori, we should expect an increase in toughness with the sintering temperature and grain size. The uncorrected K_ISB values (with a geometry factor of ψ = 1.27 for halfpenny cracks) are in the range of 14–15 MPa√m and are invariant with respect to the sintering temperature. The uncorrected K_LWN toughness values, calculated from the measured stress intensities of the surviving cracks at dummy indentations, are somewhat lower than the K_ISB values, and they also show a toughness maximum at 1350 °C.

An inspection of the fracture surfaces after an ISB test provides evidence that the zone of stable crack extension, which developed from the starter crack introduced by indentation, is very flat and not semicircular (Figure 7).

The halos of the cracks beside the indents hint at a shallow Palmqvist-type crack profile. The lateral extension is larger than the depth. Moreover, the curvature is dented in the region below the indent. A correction of the crack geometry factor, according to Newman-Raju [47], is required to compensate for the non-semicircular geometry. Dransmann [42] experimentally achieved ψ values of 1.19 and 1.08 for 4Y-TZP and 3Y-TZP by polishing out the indent and studying the crack geometry at different depths during further material removal. This procedure is, however, not applicable in the present case, as the cracks are so short that after polishing out the indent, virtually nothing will be left. The geometry factor of ψ = 0.95 was, therefore, estimated a priori from the crack geometry and contains a certain level of uncertainty.

Figure 8 shows the plots of the SIGB tests carried out with samples indented at 200 °C. Compared to cold indentation, the crack length (c₀) of the initial crack increases from 135 µm to 155 µm. Evidently, the regions beyond K_app,0 (the linear part of the curve) are well developed. At a short crack length, a deviation from the ideal behavior is observed, and the cracks grow slowly with an increasing applied-stress intensity. Due to this fading behavior, presumably caused by a subcritical crack growth, K_app,0 cannot, therefore, be determined very precisely. In the plots, K_app,0 is in the range of 4–5 MPa√m. If an extrapolation to an infinite crack length (which is unrealistic) is avoided, the maximum stress intensity measured for the longest surviving cracks is termed K_IC = K_LWN. The total crack lengths (c) in most of the cases does not exceed 0.2 µm, and the crack growth (Δc = c − c₀) is in the range of 40–80 µm. With respect to the high crack velocity, close to criticality, “catching” longer cracks would require significantly larger samples.

The threshold toughness (K_I0 = K_LWN − K_app,0) is, therefore, in the range of 3.6–5 MPa√m, which is quite close or slightly above to the typical value (3–4 MPa√m) for Y-TZP materials [55]. This result is in accordance with Imariouane’s double-torsion results [25].

The K_LWN, K_app,0, and K_I0 values corrected for a flatter crack geometry (with a geometry factor of ψ = 0.95) are shown in Figure 9. The corrected K_ISB and K_LWN values are considerably lower (~9–10 MPa√m) and are in the range of the SEVNB and double-torsion test-toughness values. Assuming K_IC = K_LWN, the measured values for the resistance to subcritical crack growths (K_I0) and the fracture toughness (K_LWN) imply that the fatigue strength values (σ_FS = σ_4pt ∙ K_I0/K_IC) are in the range of 450–500 MPa for all sintering temperatures.

According to the literature, SEVNB tests show a steep increase in toughness to a plateau value of ~8.5 MPa√m within the first 100 µm of crack extension, and double-torsion tests show toughness values of up to 9.5–10 MPa√m under fast fracture conditions [25]. A more detailed statement concerning shorter cracks is not possible with the SEVNB method.

SIGB tests with annealed and non-annealed samples were combined to construct the R-curves [42,43] in the crack length range of 0–50 µm (Figure 10). The results show that the toughness (K₀) at a zero-crack extension is 3–3.2 MPa√m, and the R-curve-dependent part of toughness is in the range of 5.5–6 MPa√m, which is in good accordance with the SIGB results. The R-curves become steeper with an increasing sintering temperature. It seems, however, that the curves have not reached the saturation level yet.

3.3. Phase Composition

Figure 11 shows an example of the XRD spectra in the monoclinic/tetragonal fingerprint range of 27–33° 2θ. The polished surface consists entirely of the tetragonal phase, and this example is characteristic for all sintering temperatures. Neither monoclinic peaks nor cubic peaks were detected. The phase composition of the fracture surfaces is predominantly monoclinic. The monoclinic fractions, determined according to [34], range from 80 to 81 vol%. This is very close to the theoretical maximum transformable content of 85–90 vol% [56]. Consequently, the transformation-zone sizes (h), calculated according to Kosmac [54], were also extremely uniform, with values of h = 3.3–3.4 µm. The calculated transformation-toughness values (ΔK_IC^T), according to McMeeking [14] (with X = 0.27), were between 4.76 and 4.96 MPa√m. The maximum values for h and ΔK_IC^T were obtained for the sample sintered at 1350 °C. The transformation-toughness values fit very well with the data from the SIGB test, assuming that K_app,0 corresponds to the toughness increments contributed by reinforcement mechanisms (in this case, only transformation toughening).

3.4. Microstructure

Figure 12 shows the SEM images of the polished and thermally etched surfaces of 1.5Y-TZP samples sintered at the four different temperatures. The microstructure is dense and homogeneous. The grain growth with an increasing sintering temperature seems very moderate.

Figure 13 shows the grain sizes determined by the linear intercept method, including a geometry correction factor of 1.56 [52]. Grain growth with an increasing sintering temperature is clearly visible. Samples sintered at 1250 °C have an average grain size of 320 nm, and the grain size increases to 450 nm at a sintering temperature of 1450 °C. Note that the given standard deviation is not the deviation in grain size (which systematically cannot be determined by the linear intercept method, as grains are sectioned at different levels) but the deviation within five measurements.

4. Discussion

The 1.5Y-TZP materials were manufactured by the axial pressing of commercially available RTP powders. Materials were characterized with respect to their microstructure, transformation characteristics, and mechanical properties. The materials show a combination of high strength and toughness within a relatively broad processing temperature field of 1250–1400 °C. This study allowed for a re-evaluation of previous publications by Matsui and Imariouane [24,25] using the same starting powder but slightly different processing conditions.

The extremely high stress-induced transformability of the materials (>80% of the phase transformation in the fracture surfaces) guarantees a high damage tolerance. Samples sintered at 1250–1300 °C showed almost no scattering in their strength data. Simultaneously, the high transformability seems to be the reason for the overestimation of fracture toughness by direct crack-length measurements. Quinn and Bradt [36] have banned direct crack-length measurements in general due to the unclear criteria for crack arrest. This popular technique is, however, very fast and easy to apply. In fact, for small samples, there are not many alternatives. In the present case, using DCM definitely cannot be recommended.

The reason for the overestimated DCM toughness values for 1.5Y-TZP [22,23,24] seems to be material specific. Indentation causes phase transformations around the indent which, due to the volume expansion of the material, prevents crack growth extremely efficiently. Hence, cracks are trapped, and regular crack patterns are seldom obtained. This phenomenon was described by Cook for 3Y-TZP materials [37]. Our experience is, however, that in case of the less-transformable 3Y-TZP, the trapping problem can be avoided, if sample preparation is gentle and automated in order to avoid the residual stress introduced by the final machining process.

This intense transformation behavior also affects the toughness measurements from residual strength measurements, if samples are notched with Vickers indents at ambient temperature. The residual strength levels in such ISB-tests may reach almost 1000 MPa and are, therefore, only marginally lower than the strength of the pristine samples. The surviving cracks of multiple indentations either grow, and the sample fails at this indent, or they are trapped, and only little crack growths are observed. In its extreme case, the crack requires such a high-stress intensity to exit the trapping zone, wherein once the region outside the zone is reached, immediate failure occurs. An indentation crack trapped by compressive stress in the transformation zone cannot be compared to a natural flaw where the transformation zone is built up during crack growth and where much more pronounced crack growths can be expected.

Indentation at elevated temperatures can help to produce more regular crack patterns. Cracks introduced by HV10 indentations at 200 °C extend 20 µm further than cracks induced at room temperature. The required applied stress to let them grow is reduced; simultaneously, the transformation around the indent is suppressed. In combination, this means a reduction in K_res, which results in regular crack patterns with four well-developed wing cracks of equal length. This facilitates the SIGB tests and introduces more statistical quality, and typically, all perpendicular crack pairs grow with an increasing applied stress and can be measured. In cold indented materials, frequently, only one crack is “running”, while the others remain trapped.

However, the temperature for hot indentation has to be chosen with care. In the case of the 1.5Y-TZP material tested in this study (which is known to be thermally vulnerable [26]), notching at 250 °C leads to the spontaneous, uncontrolled transformation of the material surrounding the indent, so that not only wing cracks but also cracks from the sides of the indents are produced. Applying this method to other materials with high transformability, therefore, should be performed with care, and the appropriate temperatures should be checked by preliminary experiments. A comparison with the results of Imariouane [25] shows that the modified indentation methodology seems valid to determine the toughness of this highly transformable material reliably. The slight reduction in toughness and strength in the TZP sintered at 1400 °C for 2 h compared to TZPs sintered at a lower temperature indicates that this material is very close to the critical grain size.

The relatively weak non-linearity of the stress–deformation curves of the 4 pt bending tests is evident. We can, however, only speculate why the slope of the stress–deformation curves do not decline continuously after triggering the phase transformation. Transformation is initiated when a certain population of large grains reaches its critical transformation stress. Transformation bands are known to exert a certain stress-shielding effect in their vicinity, which prevents the formation of new bands close to the initial ones [25]. This may restore the rigidity of the material until the stress is high enough to either lead to failure (the common case) or to the formation of new bands, while the sample survives up to a somewhat-higher stress (only observed in a single case). In the case of 3 pt bending with a localized high stress, only single-bending events occur close to the failure stress.

The calculated transformation toughness values (approximately 5 MPa√m); the K_app,0 values of the SIGB tests, which represent the R-curve-dependent part of toughness (5 MPa√m); and the R-curve-related part of toughness obtained directly from the R-curves (5.5–6 MPa√m) agree very well. It is assumed that transformation toughening is the only relevant toughening mechanism. According to Swain, the maximum toughness allowed for flaw-size-related failures in Y-TZP is ~7.5 MPa√m [13]. Consequently, depending on the K₀ value (4 MPa√m, according to Swain, and 3 MPa√m, as measured in this study), the R-curve-related toughness should not be higher than 3.5–4.5 MPa√m. The R-curve-related toughness measured in this study (5–6 MPa√m, depending on the method used) is well above this level. A transformation-related failure is confirmed by the formation of transformation bands prior to the failure. Transformation bands are, however, very narrow and start to occur at a stress slightly below the failure stress. This explains why a high strength level can be maintained. The macroscopic non-linearity of stress–strain curves was only observed in a few samples with a relatively high strength. Typically, samples fail upon reaching the critical transformation stress.

These tough 1.5Y-TZP materials offer a very narrow and reliable strength distribution and high safety in the case of catastrophic events. Under fatigue load conditions, their relatively unfavorable K_I0/K_IC ratio allows for constant stress levels that are not higher than 450–500 MPa to prevent fatigue failure. With respect to the sintering temperature and LTD resistance reported in the literature [26], it is probably advisable to not fully exploit the potential toughness and to choose a sintering temperature that is not higher than 1300 °C at a 2 h dwell. Moreover, it has to be stated that this 1.5Y-TZP is a material to be applied at ambient temperature, and exposition to elevated temperatures for a prolonged time should be avoided.

5. Conclusions

The present study shows that 1.5Y-TZP materials offer a favorable combination of a high strength of 1000 MPa and a fracture toughness of 8–10 MPa√m. The materials are, therefore, attractive in applications which require high damage tolerance in catastrophic events rather than high fatigue strength. Users aiming to apply these understabilized TZPs should be aware of the specific advantages and drawbacks of these materials, even more so as direct crack-length measurements lead to an extreme overestimation of toughness. This may lead to fatal errors in the design of components. An alternative indentation-based method, SIGB, using indents placed at elevated temperatures, is able to reliably reproduce the toughness values measured by conservative methods. Transformation toughness seems to be the only toughening mechanism active in these materials. R-curve-related toughness increments determined by different measuring protocols are in excellent agreement with calculated transformation toughness values from XRD measurements.