Novel speed sintered zirconia by microwave technology

Abstract

Objectives:

Continuous efforts have been made to hasten the zirconia densification process without compromising properties. This study evaluated the long-term structural durability of microwave speed-sintered zirconia (MWZ) relative to a conventionally sintered zirconia (CZ).

Methods:

As-machined dental 3Y-TZP discs (Ø12 × 1.2 mm) were speed sintered at 1450°C for 15 min using an industrial microwave oven, while conventional sintering was conducted in a standard dental furnace at 1530°C for 2 h. Both were followed by natural cooling. The total sintering time was 105 min for MWZ and 600 min for CZ. Groups were compared regarding density, grain size, phase composition, and fracture resistance. Structural durability was investigated employing two fatigue protocols, step-stress and dynamic fatigue.

Results:

Compared to CZ, MWZ exhibited a slightly lower density (MWZ = 5.98 g/cm³, CZ = 6.03 g/cm³), but significantly smaller grain sizes (MWZ = 0.53 ± 0.09 μm, CZ = 0.89 ± 0.10 μm), lower cubic-zirconia contents (MWZ = 15.3%, CZ = 22.7%), and poorer translucency properties (TP) (MWZ = 13 ± 1, CZ = 29 ± 0.8). However, the two materials showed similar flexural strength (MWZ = 978 ± 112 MPa, CZ = 1044 ± 161 MPa). Additionally, step-stress testing failed to capture the fatigue effect in 3Y-TZP, whereas dynamic fatigue revealed structural degradation due to moisture-assisted slow-crack-growth (SCG). Finally, MWZ possessed a slightly higher Weibull modulus (MWZ = 7.9, CZ = 6.7) but similar resistance to SCG (MWZ = 27.5, CZ = 24.1) relative to CZ.

Significance:

Dental 3Y-TZP with similar structural durability can be fabricated six-times faster by microwave than conventional sintering.

Introduction

Since the initial introduction of zirconia as a framework material for all-ceramic fixed partial dentures in the late 90s [1], zirconia has evolved into the most versatile restorative material [2]. Nowadays, its clinical indications span from porcelain-veneered systems to monolithic structures, from inlays and onlays to partial crowns, from single-unit crown and multi-unit bridges to full-arch prosthesis, and from implants to abutments [3–5]. This is attributed to its superior mechanical properties coupled with ever-improving optical properties [6]. One of the drawbacks in these materials is that soft-milled zirconia restorations need to undergo a high-temperature sintering process, which typically employs a slow heating rate (on the order of 10 °C per min) and long dwell time (on the order of hours). This is because, in conventional sintering, heat is transmitted to the surface of the prosthesis radiantly and flows into the interior via thermal conduction [7]. Due to the geometric complexity and thickness variations in dental restorations as well as the low thermal diffusivity of zirconia, uniform heat distribution may not be established quickly in conventional sintering. Therefore, achieving a uniform heat distribution in conventional sintering requires slower heating rates and a longer dwell time. A typical conventional sintering cycle for zirconia often takes 8 to 12 hours to complete.

Microwave sintering exhibits many attractive features. The volumetric nature of microwave heating allows the rapid establishment of uniform heat distribution in the specimen, which yields higher production rates, lower energy consumption, and thus a lower cost [7, 8]. However, most studies focus on the microstructural characterization and determination of bulk density of microwave sintered materials, with limited mechanical property evaluations such as elastic modulus, Vickers hardness and toughness [9–11]. A few investigated flexural strength [12, 13], but none investigated long-term strength.

When ceramic prostheses are used in dental and biomedical applications, long-term durability is of utmost concern [14]. Many laboratory studies have employed fatigue testing to evaluate the long-term strength of ceramics. Amount these testing methods, accelerated fatigue testing such as step-stress is gaining popularity [15, 16], since it is much faster than the traditional S-N curve and staircase fatigue studies. Typically, a step-stress test of at least 3 different stress profiles, and involving some twenty specimens, takes 1 – 2 weeks to complete [17, 18]; an S-N curve requires 1 – 2 months to construct [19, 20]; whereas the staircase test involves large number of specimens often on the order of n = 50 [21, 22]. However, it is not immediately apparent if such testing is able to reveal the fatigue degradation of ceramics.

In this article, a most well characterized zirconia composition (3Y-TZP) was selected and subjected to conventional and microwave sintering. The microstructure, phase assembly, translucency properties, and long-term strength properties of the microwave-sintered (MWZ) and conventionally-sintered zirconias (CZ) were analyzed and compared.

Materials and Methods

Specimen Preparation

Commercial 3Y-TZP milling pucks (Ø98 mm, Vipi Block^® Zirconn, Vipi Wieland) were obtained. From which discs (Ø15 × 1.5 mm) were CAD/CAM machined to account for a 20% sintering shrinkage. Half of the specimens were sintered in a conventional oven (Dekema Austromat 664 iSiC, Dekema Dental Keramiköfen, Freilassing, Germany) according to the manufacturer’s instructions (1530 °C, 2h), while the other half were sintered in a microwave oven (FMO-1700, INTI Furnaces, Sao Carlos, Brazil) as follows: 30 °C/min up to 1450 °C, with a dwell time of 15 minutes. In both sintering techniques, the specimens were cooled to room temperature inside the closed oven.

Materials Characterization

The zirconia microstructure was observed on finely polished and thermally etched sections. To prevent grain growth, thermal etching was carried out at a relatively low temperature (1150 °C for 20 min) and heating rate of 10 °C/min [23]. Grain size analysis was carried out in images obtained by a field emission scanning electron microscope (FEG-SEM) using a secondary electron detector (SE). At least 300 grains were measured using the linear intercept method [24]. A correction factor of 1.56 for tetrakaidecahedral grains was utilized [25].

The fractions of monoclinic, tetragonal and cubic crystalline phases were characterized by x-ray diffraction (XRD, PANalytical X’Pert) with nickel filtered Cu K_α radiation operating at 45 kV and 40 mA. Scans were performed over the 2 range 20° – 80° at a scan rate of 0.2°/min and a step size of 0.02°. The acquired diffractograms were used to quantify the tetragonal (t), cubic (c), and monoclinic (m) phases of zirconia by the Rietveld refinement method using the MAUD (Materials Analysis Using Diffraction) program [26].

Density was measured by the Archimedes principle (n = 3). The theoretical density value used for 3Y-TZP was 6.10 g/cm³ [27].

Optical Properties

Color coordinates CIE $L^{\mathrm{*}}{a}^{\mathrm{*}}{b}^{\mathrm{*}}$ [28] were measured over standard backgrounds (black $L^{\mathrm{*}}=1.8,a^{\mathrm{*}}=1.3,b^{\mathrm{*}}=-1.5$ and white $L^{\mathrm{*}}=95.7,a^{\mathrm{*}}=-1.3,b^{\mathrm{*}}=2.6$). To ensure optical continuity, a drop of Glycerol (refractive index: 1.472) was placed between the specimen and background [29].

For the translucency parameter $\left(TP\right)$ measurement, the two lateral surfaces of the sintered zirconia discs were polished to a 1-μm diamond suspension finish. The final specimen thickness was 1 mm. The $TP$ of values were determined by the color difference between the specimen on black $\left(B\right)$ and white $\left(W\right)$ backgrounds, according to Equation 1 [30]:

$$\mathrm{T}\mathrm{P}=\sqrt{({\mathrm{L}}_{\mathrm{B}}^{\mathrm{*}}-{\mathrm{L}}_{\mathrm{W}}^{\mathrm{*}}{)}^2+({\mathrm{a}}_{\mathrm{B}}^{\mathrm{*}}-{\mathrm{a}}_{\mathrm{W}}^{\mathrm{*}}{)}^2+({\mathrm{b}}_{\mathrm{B}}^{\mathrm{*}}-{\mathrm{b}}_{\mathrm{W}}^{\mathrm{*}}{)}^2}$$ (1)

where $L^{\mathrm{*}},a^{\mathrm{*}}$ and $b^{\mathrm{*}}$ refer respectively to the lightness, redness to greenness, and yellowness to blueness coordinates in the CIE color space.

Flexural Strength

Biaxial flexural strength was determined using the piston-on-3-ball (hardened steel piston and balls) method. Milled and then sintered specimens (Ø12 × 1.2 mm) were loaded monotonically at a rate of 1 mm/min until catastrophic fracture occurred. Strength was calculated using Equations 2 to 4 [31]:

$$\sigma =\frac{-0.2387F\left(X-Y\right)}{h^2}$$ (2)

$$X=\left(1+v\right)\ln {\left(\frac{c}{R}\right)}^2+\left[\frac{\left(1-v\right)}{2}\right]{\left(\frac{c}{R}\right)}^2$$ (3)

$$Y=\left(1+v\right)\left[1+\ln {\left(\frac{a}{R}\right)}^2\right]+\left(1-v\right){\left(\frac{a}{R}\right)}^2$$ (4)

where $F$ is the maximum load at fracture (N), $v$ is the Poisson’s ratio of the ceramic, $h$ is the specimen thickness (mm), $a$ is the radius of the support ring (mm), $c$ is the radius of the loaded area (mm), and $R$ is the specimen radius (mm).

Long-Term Strength

An accelerated lifetime step-stress fatigue test was used to evaluate the cyclic fatigue resistance of the studied materials. Three step-stress profiles (mild, moderate, and aggressive) were designed based on the results of the single load to fracture test [32]. For the mild profile, loading started at 200 N and progressed up to 650 N for a total of 200,000 cycles; the moderate profile started at 250 N and progressed up to 750 N for a total of 130,000 cycles; and the aggressive profile started at 300 N and progressed up to 900 N for a total of 80,000 cycles. A fourth ultra-mild profile was later defined, which started at 400 N and progressed up to 750 N for a total of 500,000 cycles. Sintered specimens were placed on a biaxial 3-ball support and cyclic loaded with a spherical WC indenter (r = 3.18 mm) on an electrodynamic fatigue testing machine (EL-3300, Enduratec Division, TA Instruments, Minnetonka, MN). Cyclic loading was applied in a sinusoidal waveform oscillating between a pre-determined maximum load and a 5-N minimum load at a frequency of 5 Hz. Specimens were distributed across the four profiles as follows: ultra-mild (n = 5), mild (n = 3), moderate (n = 3), and aggressive (n = 3).

Dynamic fatigue was also performed with the specimens placed on the same biaxial 3-ball support and loaded with a spherical WC indenter (r = 3.18 mm) to investigate the slow crack growth (SCG) behavior of the materials according to ASTM C1368 [33]. This testing method has been widely used in the engineering sector [34, 35] as well as in dental and biomedical fields [36–38]. Tests were conducted on a screw driven universal tester (Model 5566, Instron, Norwood, MA), and the loading rate was varied from 1 to 10⁻⁴ mm/min (over 5 loading rates with n = 5 at each rate). The loading rate was then converted to the stressing rate (MPa/s) for every data point. Each specimen was loaded at a fixed rate $\mathrm{d}\sigma /\mathrm{d}t$ = constant until fracture ${\sigma }_{\mathrm{F}}$_, as described in Equation 5:

$${\sigma }_F=(d\sigma /dt)t_F$$ (5)

where ${\sigma }_{\mathrm{F}}$ is the elapsed time to fracture.

Previous studies showed that the velocity exponent $N$ that describes the SCG behavior of the material can be determined using Equation 6 [39]:

$${\sigma }_F={[A(N+1)\dot{\sigma }]}^{1/(N+1)}$$ (6)

where $A$ is a load-, time-, and thickness-independent quantity, and $\dot{\sigma }=d\sigma /dt$ is the stressing rate (MPa/s).

Therefore, by plotting the ${\sigma }_F-\dot{\sigma }$ curve in logarithmic coordinates, the $N$ value can be readily derived from the slope of best-fit curves.

Substituting Equation 5 into Equation 6, we obtain the ${\sigma }_F-t_F$ relation for dynamic fatigue:

$${\sigma }_F/{\sigma }_0={\left(t_0/t_F\right)}^{1/N}$$ (7)

where ${\sigma }_0$ and $t_0$ are reference parameters relating to short-term tests.

Finally, by using Equation 7 and experimentally determined $N$ and $t_{\text{F}}$ values, flexural strength corresponding to various stressing rates can be collapsed to fast fracture or short-term flexural strength, which can be related to the short-term ceramic strength, ${\sigma }_0$, obtained from Equation 4.

Weibull Analysis

To provide reliable prediction of fracture resistance of ceramics, the Weibull failure probabilities should be determined. The Weibull failure probability is described by the Weibull modulus, $m$. A higher $m$ value indicates a smaller scatter in measured properties.

In terms of critical flexural stress (strength) ${\sigma }_F$ of zirconia discs, the Weibull failure probability $P$ can be defined as:

$$P=1-\mathrm{e}\mathrm{x}\mathrm{p}\left[-{\left({\sigma }_F/{\sigma }_0\right)}^m\right]$$ (8)

where ${\sigma }_0$ is a scaling stress or the short-term flexural strength. For a data set of critical stresses, cumulative probabilities are calculated by ranking values in ascending order and evaluating corresponding ${\sigma }_F$ values. A plot of $ln(ln(1/(1-P)))$ against $ln{\sigma }_{\mathrm{F}}$ gives a straight line with slope $m$.

Statistical Analysis

Statistical analyses were conducted to compare the inert strength, SCG velocity exponent, and Weibull modulus between MWZ and CZ. We used an α level of 0.05 for all statistical tests.

The inert strength data of both MWZ and CZ passed the Shapiro-Wilk Normality test (p = 0.383) and the Equal Variance test (p = 0.233), and were statistically analyzed using t-test.

Both the SCG velocity exponent $N$ and Weibull modulus $m$ are derived from the slope of a linear regression line. For example, the SCG velocity exponent $N$ can be expressed as follows: $log{\sigma }_F=[1/(N+1)]log\dot{\sigma }+logD$. The Weibull modulus $m$ is expressed as following: $ln(ln(1/(1-P)))=m\left(ln{\sigma }_F-ln{\sigma }_0\right)$. Therefore, the statistical test of the $N$ and $m$ values can be carried out by critically examining the difference between the slopes from two independent samples using both pooled and unpooled error variances according to Howell [40].

Results

Table 1 provides an overview of the study results. Details supporting the derivation of these key results are presented in the following figures.

Table 1:

Overview of the study results comparing the properties of zirconia speed sintered by microwave versus conventionally sintered

	MWZ	CZ
Grain size (μm)	0.53 ± 0.09	0.89 ± 0.10
Phase fractions (%)
t-ZrO2	84.2	72.3
c-ZrO2	15.3	22.7
m-ZrO2	0.5	5
Relative density (%)	98	99
Translucency parameter, TP	13 (1)	29 (0.8)
Flexural strength, σ (MPa)	978 (112)	1044 (161)
Weibull modulus, m	7.9	6.7
Velocity exponent, N	27.5	24.1

The microstructure, phase assembly, and translucency properties of the zirconia ceramics investigated are shown in Figure 1. FE-SEM images of thermally etched surfaces demonstrated that the average grain size was smaller for MWZ, which also contained sparse, small pores (Figure 1a). X-ray diffraction analysis revealed slight differences between the two sintering techniques (Figure 1b). Phase quantification identified higher c-ZrO₂ content in CZ. It also revealed a trace amount of m-ZrO₂ in CZ. This is due mainly to non-uniform dispersion of yttria additives in the starting powder, leading to the localized depletion of yttrium in the specimen. Upon cooling, such an yttrium depletion could fail to stabilize zirconia and result in the formation of m-ZrO₂. The density of the CZ (6.03 ± 0.02 g/cm³) was slightly higher (p < 0.001) than that of MWZ (5.98 ± 0.03 g/cm³). The $TP$ values measured for the two sintering methods were also significantly different (p < 0.001), with CZ being more translucent than MWZ (Figure 1c).

Figure 1:

Characterization results for microwave- and conventionally-sintered zirconias. (a) Micrographs produced by SEM, (b) XRD spectra showing the monoclinic peaks with a * symbol and the cubic peak with a ▼ symbol, and (c) digital photograph illustrating the materials’ translucency.

Regarding the mechanical properties of the zirconias tested, the two sintering techniques yielded similar fast fracture or inert flexural strength: MWZ = 978 ±112 MPa, CZ = 1044 ±161 MPa. The t-test showed that there is no difference in the mean strength values between the two groups (p = 0.188).

Figure 2 depicts the 4 step-stress profiles utilized in the current study, namely aggressive (Agg), moderate (Mod), mild (Mil), and ultra-mild (U-Mil). Each load level and its corresponding stress level are plotted as a function of number of cycles and the equivalent testing duration. The corresponding stress level was estimated by using the targeted specimen thickness h = 1.2 mm along with a mean specimen radius R = 0.71 mm. It is important to note that most of the specimens were fractured around 540 – 760 N (see the grey shaded band), regardless of the loading profile and number of cycles.

Figure 2:

Illustrations of the load (or stress) increase with number of cycles (or test durations) in each step-stress fatigue profiles: aggressive (Agg), moderate (Mod), mild (Mil), and ultra-mild (U-Mil). The load range where most of the samples fractured is depicted by the shaded band. The 5 stress rates of the dynamic test are also included for comparison: 1, 10⁻¹, 10⁻², 10⁻³, and 10⁻⁴ mm/min, indicated by grey lines labelled as I, II, III, IV, V, respectively.

For reference, the five loading rates employed in the dynamic fatigue tests are also plotted (grey lines). They were 1 (I), 10⁻¹ (II), 10⁻² (III), 10⁻³ (IV), and 10⁻⁴ (V) mm/min, which corresponded to 90, 9, 0.9, 0.09, and 0.009 N/s or 100, 10, 1, 0.1, 0.01 MPa/s, respectively. As can be seen, the four different step-stress profiles can also be perceived as four varying loading rates, for which stress stepped up as a function of test durations. The effective loading rate of the three faster step-stress profiles (aggressive, moderate, and mild) lay between 0.09 and 0.009 N/s.

Figure 3 shows results of the dynamic fatigue tests as flexural strength versus the stressing rate in logarithmic coordinates for both MWZ (a) and CZ (b). Solid lines are regression power-law fits to Equation 6. Grey curves are 95% confidence bounds to the data set. Crack velocity exponents $N$ derived from the inverse of slopes of the fitted lines show $N=27.5$ for MWZ and $N=24.1$ for CZ. It is evident that the structural degradation due to moisture assisted SCG yields strength reduction down to ~700 MPa for both groups. Indeed, statistical analysis using the method for comparing the slopes of two independent samples [40] gives pooled and upooled solutions of p = 0.62767 and p = 0.62773, respectively. These findings suggest that there is no statistical difference between the $N$ values for MWZ and CZ.

Figure 3:

Flexural strength versus stressing rate for (a) MWZ and (b) CZ under dynamic fatigue. Solid lines are logarithmic regression fits to raw data, in accordance with slow-crack-growth (SCG) analysis. Grey curves represent 95% confidence bounds.

The dynamic strength data (Dyn) in Figure 3 are replotted in Figure 4a as a function of test duration $t_{\mathrm{F}}$, using $t_{\mathrm{F}}=\sigma /\dot{\sigma }$. For comparison, step-stress strength data (SS, grey filled symbols) are also included. As can be seen, despite the step-stress strength data being concentrated between 6.5 × 10³ and 2.5 × 10⁵ s, they nevertheless overlap with the dynamic strength data.

Figure 4:

(a) Replot of dynamic fatigue data (Dyn, open symbols) in Figure 3 as strength versus test duration. Step-stress strength data (SS, grey filled symbols) are also plotted for comparison. (b) Weibull plots for dynamic strength data of both zirconia materials. Data were reduced to fast fracture inert strength using Equation 7.

Weibull plots for dynamic strength data of MWZ and CZ are shown in Figure 4b. Flexural strength data corresponding to different load-rates were reduced to short-term flexural strength (inert strength) at the 1-mm/min loading rate by using Equation 7. Such an approach is based on a probabilistic effective inert strength model derived from the SCG concept in conjunction with the Weibull distribution, where the experimental strength data obtained from various loading rates can be conjointly evaluated after converting the data to an effective inert strength [34, 39]. Statistical analysis that tests the difference between the slopes from two independent samples [40] gives pooled and upooled solutions of p = 0.00379 and p = 0.00267, respectively. These findings suggest that MWZ exhibits a higher Weibull modulus ($m=7.9$) than that ($m=6.7$) of CZ. In addition, we note that both Weibull curves have a similar kink at higher loads. This bears further investigation.

Discussion

The results presented in this study demonstrate microwave technology as an efficient method to sinter 3Y-TZP. It is 6 times faster than conventional sintering, yet produces materials with similar physical (bulk density) and mechanical (flexural strength) properties but with significantly lower translucency. In clinical applications, the strong but less translucent 3Y-TZP ceramics are indicated for frameworks for single-unit and multi-unit restorations as well as ceramic implants and abutments [41]. Thus, the relatively low translucency of MWZ is not going to be a significant disadvantage, while gaining a factor of 6 reductions in fabrication time while conserving energy. In addition, microwave sintering offers many other attractive features, including uniform heat distribution, higher densification rates, a better-controlled microstructure, and shortened sintering time. This, in turn, leads to higher production rates, lower energy consumption, and thus a lower cost. However, microwave sintering is still a technology in progress and sintering protocols for zirconia are yet to be optimized to achieve better density and translucency. Furthermore, although its dwell time for densification (~15 mins) has drastically reduced from that of conventional sintering (~2 h), its heating rate (30 °C/min) is still relatively low compared to some recent speed sintering protocols (~100 °C/min) [23, 42].

Our findings confirm the susceptibility of zirconium oxide ceramics to strength degradation by moisture-assisted SCG [43]. A fracture mechanics analysis incorporating a crack velocity equation provides the basis for quantifying the rate dependence of the flexural strength. MWZ and CZ possess similar values of velocity exponent parameter $N$, suggesting that the two materials should exhibit similar long-term durability. Although both MWZ and CZ have an identical starting composition, the characteristics of their respective sintered specimen, i.e. grain size, phase assembly, and porosity determine the Weibull modulus value. This appears to be borne out by the higher $m$ value for MWZ, as an indicator of a more uniform microstructure.

The step-stress testing showed both zirconia fracture at around 540 – 760 N, or 550 – 780 MPa, irrespective of loading profiles (ultra-mild or aggressive) and number of cycles. However, when the number of cycles to fracture are converted to the effective test durations, all step-stress data clustered between 10⁴ and 10⁵ seconds, which is clearly not sufficient to determine the strength degradation dependence on stress rate or test duration (Figure 4a). This is because the effective loading rate or stressing rate in the three faster step-stress profiles (aggressive, moderate, and mild) lie between the two slowest loading rates of dynamic testing, suggesting that the current step-stress profiles are not capable of determining the fast fracture strength, and thus are unable to detect strength degradation due to moisture-assisted SCG. Unfortunately, given the nature of the step-stress profiles, it would be very challenging to fracture a specimen within tens of seconds. In order to cover a time span of 5 decades as recommended for dynamic fatigue testing [35], one would need to extend the test duration to 10⁶ or 10⁷ secs (i.e., 10 – 100 days). This defeats the very principle of the step-stress approach, which is designed to accelerate fatigue testing.

To reiterate, when conducting any accelerated laboratory fatigue study, care must be taken in designing appropriate testing protocols to cover a wide range of stressing rates in order to reveal the fatigue behavior. In addition, when the step-stress data are plotted alongside the dynamic data, they overlap each other, suggesting that in as-sintered specimens flexural fatigue fracture is driven predominantly by SCG without significant mechanical degradation. This has been observed previously for ceramics with polished surfaces [44], but not with surfaces which contain microdamage induced from sandblasting [45] as well as sharp [46] or blunt [44] contacts.

Conclusions

Principal conclusions from this study are as follows:

• Microwave sintered zirconia can be prepared 6 times faster than conventional sintering without compromising structural durability.

• Degradation in flexural fatigue of as-sintered zirconia is due predominantly to moisture-assisted SCG.

• MWZ possesses higher Weibull modulus relative to CZ.

• MWZ exhibits similar resistance to SCG relative to CZ.

• When using accelerated fatigue testing to reveal the fatigue behavior, care must be taken in designing appropriate testing protocols to cover a wide range of stress-rates.

Highlights.

• MWZ can be prepared 6 times faster than CZ while preserving structural durability.

• Degradation in flexural fatigue of as-sintered zirconia is due primarily to SCG.

• MWZ possesses a slightly higher Weibull modulus than CZ.

• MWZ and CZ exhibit similar resistance to SCG.

Accelerated fatigue testing protocols must cover a wide range of stress-rates.