Abstract
Objective
Processing parameters (powder granulation, compaction, debinding, greenbody shaping, sintering) and post-sinter rough, even fine grinding are influencing the final mechanical properties of 3Y-TZP. The hypothesis of this study was that post-sinter hot isostatic pressing (post-HIP) would be beneficial for improving reliability and strength of both sintered and coarse ground sintered zirconia by closing or reducing surface and/or small volume defects.
Methods
75 sintered bars of an experimental 3Y-TZP (3 mm × 4 mm × 45 mm) with chamfered edges and 15 μm diamond surface finish were provided by the manufacturer (Ivoclar Vivadent) and randomly distributed in five groups of N = 15 each. G1 served as control (as received); G2 was post-HIPed at 1400 °C and G3 at 1350 °C, both using a pressure of 195 MPa in Ar for 1 h; G4 was coarse ground with 120 μm diamond disk grain size; G5 was ground 120 μm and post-HIPed at 1350 °C at 195 MPa, 1 h in Ar. The specimens were fractured in air in 4 point-bending. Weibull characteristic strength ( σ 0 ) in MPa, m parameter (reliability) and confidence intervals (CI) at 90% confidence level are reported. Identification of the critical flaw was performed by SEM on the fractured surface of all specimens and XRD performed in all groups.
Results
G1: σ 0 = 973 (932–1016), m = 10.6 (7.45–15.1); G2: σ 0 = 930 (871–995), m = 6.9 (4.87–9.9); G3: σ 0 = 898 (848–952), m = 7.94 (5.6–11.4); G4: σ 0 = 921 (857–991), m = 6.35 (4.48–9.11); G5: σ 0 = 881 (847–918), m = 11.4 (8.03–16.3). G5 had a significantly lower σ 0 than G1. No significant differences were seen in the reliability ( m ) among the groups. Fractography revealed critical intrinsic subsurface flaws of 10–60 μm present in all groups resulting from the processing parameters. No “healing” (i.e. closing of defects by densification) resulted after post-HIP. Grinding sintered zirconia with 120 μm diamond disks induced radial cracks of 10–20 μm and an important pseudo-cubic phase transformation (56 wt%) that was not completely removed after post-HIP. Post-HIP increased slightly the relative density by 0.1% but without improving the strength and reliability.
Significance
Post-HIP was not efficient in closing large (10–60 μm) subsurface (volume) processing defects.
1
Introduction
Yttria stabilized tetragonal zirconia polycrystal (Y-TZP) has become an alternative high-toughness all-ceramic core material for fixed-partial dental restorations because of its excellent mechanical properties (i.e., bend strength and toughness). The stress-induced phase transformation of Y-TZP from the tetragonal crystalline structure to the more voluminous monoclinic structure helps to prevent crack propagation and contributes to the strength and toughness of Y-TZP ceramics. Nevertheless, the final properties of zirconia will heavily depend on both the production processing steps as well as the introduction of surface damage caused by grinding . Processing related defects on 3Y-TZP ceramics were shown to lowering the fatigue limits, fracture resistance and reliability . Typical processing related defects are porous regions or crack-like voids during powder compaction, extraction of the binders or solid-state sintering as shown using fractography on failed ceramic parts . To overcome some of these processing defects, hot isostatic pressing (HIP) is often used in the ceramic industry as an additional step after sintering to densify the zirconia using high temperature (∼100–200 °C below the zirconia’s sintering temperature) and isostatic gas pressure (argon). It leads to various mass transport mechanisms such as grain boundary sliding, plastic deformation and diffusion controlled creep to improve densification, crack healing and pore shrinkage in Y-TZP without shape changes. This HIP procedure after sintering is called post-sinter HIP or post-HIP and has been shown to increase the fracture resistance of 3Y-TZP root dental posts . Mechanical properties such as modulus of elasticity and fracture toughness were also enhanced when using post-HIP on zirconia-toughened alumina ceramics as well as the fracture strength and thermal shock resistance .
Another detrimental step to ceramic materials is the introduction of surface flaws by grinding after sintering . The use of medium to coarse diamond burs (grain size 50–120 μm) on sintered zirconia have shown to introduce surface flaws extending radially up to 50 μm thus limiting its strength, reliability and survival rate under cyclic loading . Even finer diamond burs of 30–40 μm grain size may introduce subsurface damage with horizontal cracks up to 20 μm . Classic fracture mechanics applicable to brittle materials show that the strength of ground zirconia is mainly determined by the critical defect size to initiate failure. Thus, for a Y-TZP with a fracture toughness of 5 MPa√m, a critical flaw size of 50 μm and a Y factor of 1.2, the failure stress will be of 600 MPa, whereas it will be of 946 MPa for a flaw size of 20 μm. Considering the existence of small (2–15 μm) CAD-CAM machining flaws of sintered Y-TZP and additionally possible introduction of manual grinding flaws (up to 50 μm) with dental burs due to the habit of reshaping sintered zirconia frameworks in the laboratories prior to the application of a veneering ceramic, there is a need to reduce the flaw population within the zirconia framework. Crack healing of surface flaws has been described by using heat treatment (annealing) on thermally shocked alumina and Vickers indented zirconia/alumina composites . The crack-healing mechanism has been shown to be related to the evolution of pore morphology during the heat treatment and rearrangement of grains caused by the monoclinic to tetragonal phase transformation in the crack-opening process . Post-HIP and crack healing operations have also been described to improve survivability of silicon nitride/silicon-carbide composite cutting tools .
Considering that CAD-CAM made dental zirconia frameworks are often reshaped manually post sintering in the dental technician’s laboratory with possible introduction of critical grinding surface flaws adding to existing processing defects, it is the hypothesis of this study that post-sinter HIP will improve strength and reliability of the zirconia ceramic by reducing in size or even eliminating internal processing defects and operate partial or even full crack healing on surface grinding flaws. The option of healing critical flaws resulting from processing and/or reshaping by rough grinding sintered CAD-CAM zirconia frameworks will provide a significant quality insurance for laboratories and dentists, with a mechanically improved framework and thus increased life expectance.
The aims of this research were to:
- a)
determine the effect of post-sinter HIP treatment on the mechanical properties (characteristic strength and reliability) of a 3Y-TZP and after rough surface grinding with a 120 μm diamond grain size disk.
- b)
identify by fractography the critical flaw responsible for failure after strength testing and what effect post-sinter HIP had on the critical flaw morphology.
- c)
characterize the material’s microstructure and density after sintering and post-sinter HIP at two different HIP temperatures.
- d)
quantify the radial crack damage from 120 μm diamond grinding using the bonded interface technique.
- e)
perform a semi-quantitative phase analysis by comparing the integrated intensities of the XRD reflections.
2
Materials and methods
2.1
Specimen preparation
A 3Y-TZP-A powder containing a small dispersion of Al 2 O 3 (<0.5 wt%) and without any additional granule modification was cold isostatically pressed (CIP) to rods of a length of 240 mm and a diameter of 22 mm. The pressing pressure was over 3000 bar and the maximum pressure was held for 30 s. After CIP, process cylinders of a diameter of 8.0 ± 0.5 mm and a length of 56 ± 1 mm were dry-turned by green machining on a lathe-machine. The cylinders were then dry-milled to the following bend bars dimensions of 56 mm × 5 mm × 3.7 mm. The debinding and final sintering was done in one firing cycle using a maximum temperature of 1500 °C for 2 h. After sintering, a final surface finish of 15 μm diamond grain size according to ISO 6872:2008 was performed to achieve the necessary final sample geometry of 45 mm × 4 mm × 3 mm and chamfered edges. A total of 75 bend bars were delivered as such (labeled “as received”) by the manufacturer (Ivoclar-Vivadent, Schaan, Liechtenstein) and distributed into five groups as shown in Table 1 .
| Groups of 3Y-TZP | N |
|---|---|
| G1: Control (as received) | 15 |
| G2: Control + HIP 1400 °C | 15 |
| G3: Control + HIP 1350 °C | 15 |
| G4: Control + mirror polish + ground 120 μm | 15 |
| G5: Control + mirror polish + ground 120 μm + HIP 1350 °C | 15 |
2.2
Post-sinter HIP procedure
Bend bars of G2, G3 and G5 were post-sinter treated by hot isostatic pressing (HIP) in a specific furnace (EPSInt HIP 2000, Temse, Belgium) at 1400 °C (G2) or 1350 °C (G3 and G5) at a pressure of 195 MPa in argon for 1 h. The bars to be HIPed were placed in a coarse-grained Al 2 O 3 powder-bed inside an Al 2 O 3 -crucible covered by a lid as the heating elements of the HIP are made out of carbon. To remove gray coloring from the post-sinter HIP process the bars were additionally heat treated at 1200 °C for 1 h in air (Carbolite HTF1700, Ubstadt-Weiher, Germany).
The two post-sinter HIP temperatures 1400 °C for G2 and 1350 °C for G3 were selected based on previously gained experience of the authors on similar Y-TZP and on temperatures used in the literature . The decision of which HIP temperature would be kept throughout the study was based on whether or not undesired grain growth would occur.
2.3
Mechanical testing
75 bend bars (G1–5) were tested in a 4 point-bending configuration with an inner span of 20 mm and an outer span of 40 mm. The cross-head speed was selected so that the time till failure of the bend bars occurred in a time window of 5–15 s. For the test machine and jig used in that study this resulted in a cross-head speed of 4 mm/min. Data is described by the Weibull characteristic strength ( σ 0 ) and reliability parameter ( m ) using the Maximum Likelihood Estimate (MLE), a confidence level of 90% and a correction factor for bias for the m parameter. That correction is necessary as a Weibull modulus estimated by MLE has a bias which gives an overestimate of the true Weibull modulus m . The correction factor to unbiase m was taken from Annex E of EN 843-5:2006.
2.4
Microstructure and density assessment
The microstructure was determined for G1 (control as received), G2 (HIP 1400 °C) and G3 (HIP 1350 °C) on mirror polished and thermally etched surfaces. The specimen was thermally etched at 1420 °C for 15 min. Scanning Electron Microscope (SEM) images were taken at 30,000× magnification and the average grain sizes were measured using the interceptive line technique according to DIN EN 623-3:2003 applying an additional correction factor of 1.56 for the final calculation of the mean value. The number of intersections was counted using a semiautomatic software tool (Lince 2.3.1d, Technische Universität Darmstadt), where different types of intersections were distinguished such as “normal”, “tangential”, “triple-point”, “end of line at grain boundary” or “within a grain”. Standardized line-patterns were used yielding a number of intersections over 100.
The density was measured on fractured specimens by means of the sample immersion technique which is based on the Archimedes’ principle. Due to the low weight of the specimen and the higher wettability CCl 4 was used as immersion medium to determine the ascending force and the volume of the specimen.
The Archimedes density measurements were performed on 3 fractured specimens of G1 (control as received), G3 (HIP 1350 °C), G4 (Ground 120 μm) and G5 (Ground 120 μm and HIP 1350°).
2.5
Rough grinding and damage assessment
Bars in G4 and G5 were first mirror polished (1 μm) for obtaining an initial “defect-free” surface on a Struers RotoPol-22 turntable before being ground under water for 3 min at 300 rpm on a 120 μm diamond grain size disk (Presi RefleX, Presi Sàrl, Le Locle, Switzerland).
Two additional specimens were used for assessment of the grinding damage using the Bonded Interface Configuration in which two sides perpendicular to each other are mirror polished. The two specimens are then bonded together on one of their mirror polished sides while the resultant top mirror polished surfaces are ground under water with a 120 μm diamond disk for 3 min at 300 rpm. After debonding the specimens, the extent of subsurface damage is measured under SEM on the polished side perpendicular to the ground surface.
2.6
X-ray diffraction (XRD)
X-ray diffraction (XRD) was performed on PANalytical X’Pert PRO diffractometer in Bragg–Brentano ( θ –2 θ ) configuration. The incident X-rays had a wavelength of 1.540598 Å (Cu-Kα 1 ). The diffraction patterns were scanned over the angular range 20° < 2 θ < 100° with an angular step interval of 0.0167°. The samples were adhered using amorphous organic grease onto a vicinal-cut Si single-crystal plate. The crystalline phase identification from the phase-check data was performed using the JCPDS reference-card database and the ‘Search-Match’ utility of the software PANalytical X’Pert HighScore Plus. Monoclinic phase quantification was done by using the ‘matrix method’, as described in Garvie and Nicholson . The areas of the reflections were determined through modeling the observed reflections with linear background and asymmetric Lorentzian functions. Reflections were fitted into a diffraction pattern with the least-squares utility provided by the software WinXAS 3.2.The lattice parameters were determined with the Le Bail method using the FullProf software program . The profiles were fitted using Thompson-Cox-Hastings pseudo-Voigt function . The instrumental contribution for peak broadening was estimated with the measurement of the standard reference sample CeO 2 (NIST SRM674b).
2
Materials and methods
2.1
Specimen preparation
A 3Y-TZP-A powder containing a small dispersion of Al 2 O 3 (<0.5 wt%) and without any additional granule modification was cold isostatically pressed (CIP) to rods of a length of 240 mm and a diameter of 22 mm. The pressing pressure was over 3000 bar and the maximum pressure was held for 30 s. After CIP, process cylinders of a diameter of 8.0 ± 0.5 mm and a length of 56 ± 1 mm were dry-turned by green machining on a lathe-machine. The cylinders were then dry-milled to the following bend bars dimensions of 56 mm × 5 mm × 3.7 mm. The debinding and final sintering was done in one firing cycle using a maximum temperature of 1500 °C for 2 h. After sintering, a final surface finish of 15 μm diamond grain size according to ISO 6872:2008 was performed to achieve the necessary final sample geometry of 45 mm × 4 mm × 3 mm and chamfered edges. A total of 75 bend bars were delivered as such (labeled “as received”) by the manufacturer (Ivoclar-Vivadent, Schaan, Liechtenstein) and distributed into five groups as shown in Table 1 .
| Groups of 3Y-TZP | N |
|---|---|
| G1: Control (as received) | 15 |
| G2: Control + HIP 1400 °C | 15 |
| G3: Control + HIP 1350 °C | 15 |
| G4: Control + mirror polish + ground 120 μm | 15 |
| G5: Control + mirror polish + ground 120 μm + HIP 1350 °C | 15 |
2.2
Post-sinter HIP procedure
Bend bars of G2, G3 and G5 were post-sinter treated by hot isostatic pressing (HIP) in a specific furnace (EPSInt HIP 2000, Temse, Belgium) at 1400 °C (G2) or 1350 °C (G3 and G5) at a pressure of 195 MPa in argon for 1 h. The bars to be HIPed were placed in a coarse-grained Al 2 O 3 powder-bed inside an Al 2 O 3 -crucible covered by a lid as the heating elements of the HIP are made out of carbon. To remove gray coloring from the post-sinter HIP process the bars were additionally heat treated at 1200 °C for 1 h in air (Carbolite HTF1700, Ubstadt-Weiher, Germany).
The two post-sinter HIP temperatures 1400 °C for G2 and 1350 °C for G3 were selected based on previously gained experience of the authors on similar Y-TZP and on temperatures used in the literature . The decision of which HIP temperature would be kept throughout the study was based on whether or not undesired grain growth would occur.
2.3
Mechanical testing
75 bend bars (G1–5) were tested in a 4 point-bending configuration with an inner span of 20 mm and an outer span of 40 mm. The cross-head speed was selected so that the time till failure of the bend bars occurred in a time window of 5–15 s. For the test machine and jig used in that study this resulted in a cross-head speed of 4 mm/min. Data is described by the Weibull characteristic strength ( σ 0 ) and reliability parameter ( m ) using the Maximum Likelihood Estimate (MLE), a confidence level of 90% and a correction factor for bias for the m parameter. That correction is necessary as a Weibull modulus estimated by MLE has a bias which gives an overestimate of the true Weibull modulus m . The correction factor to unbiase m was taken from Annex E of EN 843-5:2006.
2.4
Microstructure and density assessment
The microstructure was determined for G1 (control as received), G2 (HIP 1400 °C) and G3 (HIP 1350 °C) on mirror polished and thermally etched surfaces. The specimen was thermally etched at 1420 °C for 15 min. Scanning Electron Microscope (SEM) images were taken at 30,000× magnification and the average grain sizes were measured using the interceptive line technique according to DIN EN 623-3:2003 applying an additional correction factor of 1.56 for the final calculation of the mean value. The number of intersections was counted using a semiautomatic software tool (Lince 2.3.1d, Technische Universität Darmstadt), where different types of intersections were distinguished such as “normal”, “tangential”, “triple-point”, “end of line at grain boundary” or “within a grain”. Standardized line-patterns were used yielding a number of intersections over 100.
The density was measured on fractured specimens by means of the sample immersion technique which is based on the Archimedes’ principle. Due to the low weight of the specimen and the higher wettability CCl 4 was used as immersion medium to determine the ascending force and the volume of the specimen.
The Archimedes density measurements were performed on 3 fractured specimens of G1 (control as received), G3 (HIP 1350 °C), G4 (Ground 120 μm) and G5 (Ground 120 μm and HIP 1350°).
2.5
Rough grinding and damage assessment
Bars in G4 and G5 were first mirror polished (1 μm) for obtaining an initial “defect-free” surface on a Struers RotoPol-22 turntable before being ground under water for 3 min at 300 rpm on a 120 μm diamond grain size disk (Presi RefleX, Presi Sàrl, Le Locle, Switzerland).
Two additional specimens were used for assessment of the grinding damage using the Bonded Interface Configuration in which two sides perpendicular to each other are mirror polished. The two specimens are then bonded together on one of their mirror polished sides while the resultant top mirror polished surfaces are ground under water with a 120 μm diamond disk for 3 min at 300 rpm. After debonding the specimens, the extent of subsurface damage is measured under SEM on the polished side perpendicular to the ground surface.
2.6
X-ray diffraction (XRD)
X-ray diffraction (XRD) was performed on PANalytical X’Pert PRO diffractometer in Bragg–Brentano ( θ –2 θ ) configuration. The incident X-rays had a wavelength of 1.540598 Å (Cu-Kα 1 ). The diffraction patterns were scanned over the angular range 20° < 2 θ < 100° with an angular step interval of 0.0167°. The samples were adhered using amorphous organic grease onto a vicinal-cut Si single-crystal plate. The crystalline phase identification from the phase-check data was performed using the JCPDS reference-card database and the ‘Search-Match’ utility of the software PANalytical X’Pert HighScore Plus. Monoclinic phase quantification was done by using the ‘matrix method’, as described in Garvie and Nicholson . The areas of the reflections were determined through modeling the observed reflections with linear background and asymmetric Lorentzian functions. Reflections were fitted into a diffraction pattern with the least-squares utility provided by the software WinXAS 3.2.The lattice parameters were determined with the Le Bail method using the FullProf software program . The profiles were fitted using Thompson-Cox-Hastings pseudo-Voigt function . The instrumental contribution for peak broadening was estimated with the measurement of the standard reference sample CeO 2 (NIST SRM674b).
3
Results
3.1
Microstructure and density assessment
The interceptive line technique applied to calibrated images and identical line pattern lead to an average grain size for G1 (control, as received) of 476 ± 73 nm, G2 (HIP 1400 °C) of 499 ± 68 nm and G3 (HIP 1350 °C) of 475 ± 65 nm. A small increase in grain size of 25 nm is noted for the 1400 °C HIP specimen ( Fig. 1 ).
Density measurements using Archimedes are summarized in the Table 2 . HIP at 1350 °C showed an increase in the relative density of 0.1% reaching 99.7% of the theoretical density of 6.09 g/cm 3 .
| Group | N | Density (g/cm 3 ) | Relative density (% TD) |
|---|---|---|---|
| G1 (as received) | 3 | 6.067 ± 0.002 | 99.62 ± 0.04 |
| G3 (HIP 1350 °C) | 3 | 6.073 ± 0.001 | 99.72 ± 0.02 |
| G4 (Ground 120 μm) | 3 | 6.066 ± 0.001 | 99.60 ± 0.02 |
| G5 (Ground 120 μm + HIP 1350 °C) | 3 | 6.074 ± 0.002 | 99.74 ± 0.03 |
3.2
Grinding damage assessment
Damage from rough grinding with 120 μm diamond disk is documented in the following image ( Fig. 2 ). The grinding damage in form of scalloped chips into the bulk of the zirconia corresponds to radial cracks in the order of magnitude of 10–20 μm in depth.
3.3
Mechanical testing (4 point-bending)
The 4 point-bending Weibull characteristic strength ( σ 0 ) and reliability m parameter with respective confidence intervals (CI) calculated with a 90% confidence level are summarized in Table 3 . A significant difference in the characteristic strength was seen only between G1 (as received) and G5 (ground 120 μm + HIP 1350°). With only 15 specimens per group, no significant differences in the reliability are shown after rough grinding (120 μm) although the m value is low ( m = 6.35).
| Experimental groups (3Y-TZP) | N | Weibull ( σ 0 ) + CI (MPa) | m (reliability) + CI |
|---|---|---|---|
| G1: Ctr (as received) | 15 | 973 (932–1016) a | 10.6 (7.45–15.1) a |
| G2: Ctr + HIP 1400 °C | 15 | 930 (871–995) a,b | 6.9 (4.87–9.9) a |
| G3: Ctr + HIP 1350 °C | 15 | 898 (848–952) a,b | 7.94 (5.6–11.4) a |
| G4: Ctr + mirror polish + ground 120 μm | 15 | 921 (857–991) a,b | 6.35 (4.48–9.11) a |
| G5: G4 + HIP 1350 °C | 15 | 881 (847–918) b | 11.4 (8.03–16.3) a |
The effect of post-sinter HIP at 1400 °C (G2) and 1350 °C (G3) compared to the control (G1) is shown graphically in Fig. 3 representing the Weibull cumulative probability of failure distribution (left) and reliability (right). Color labels are blue for G1 (control), red for G2 (1400 °C), and black for G3 (1350 °C).
The effect of rough grinding (G4, green) and post grinding HIP (G5, black) compared to the control (G1, blue) is shown in the Weibull cumulative probability of failure distribution of Fig. 4 . A significant lowering in the σ 0 characteristic strength at 63% (shift to the left) is shown for G5 compared to G1. Rough grinding lowered the reliability ( m = 6.35) but not significantly.
The fractographic analysis of the failed specimens showed the same dominant type of flaw in all of the tested groups ( Fig. 5 ) which is a porous region near the surface. These processing (i.e., pressing and sintering) subsurface flaws were in the order of 10–60 μm. Fig. 5 a and b illustrates porous regions as failure origins in the control group G1 (as received). Post-HIP at either 1400 °C (G2) ( Fig. 5 c) or 1350 °C (G3) ( Fig. 5 d) was not able to close any of the existing critical processing flaws from the as-received state. In Group 4 ( Fig. 5 e), rough grinding (120 μm) of the tensile surface did not modify critical pressing and sintering flaws (15–20 μm) related to the original processing of the bend bars. A failure relevant flaw of a specimen ground 120 μm followed by post-HIP 1350 °C (G5) is shown in Fig. 5 f. Post-HIP smoothened somehow the grinding marks compared to the image in Fig. 5 e but apparently had no ability in closing the critical subsurface defect.
3.4
Structural transformation (XRD)
Enlarged XRD patterns within the range 25° < 2 θ < 40° for the different groups are shown in Fig. 6 . The main crystalline phase providing a sharp main reflection at around 30° in 2 θ was indexed to a tetragonal structure (space group P42/nmc ) with a diffraction pattern similar to the Zr 0.94 Y 0.06 O 1.88 (JCPDS-PDF: 01-089-9068) reference pattern. Broad reflections are observable within 2 θ < 30° in the sample G1 (as received) and G4 (Ground 120 μm), matching with the Zr 0.72 Y 0.28 O 1.862 (JCPDS-PDF: 01-077-2112) reference pattern presenting cubic crystal structure (space group: Fm-3m ). However, as discussed below, several crystal symmetries slightly distorted from cubic may explain these broad reflections and the phase(s) behind them will therefore be referred below commonly as the pseudo-cubic phase. Additionally, the reflection at 2 θ = 28.2°, matching with the monoclinic (space group: P2/c ) ZrO 2 phase (JCPDS: 00-007-0343, baddeleyite), was observed for the three samples G1, G4, G5 and was attributed to the monoclinic phase ( P2/c ). For the samples G2 (HIP 1400 °C) and G3 (HIP 1350 °C), however, only the tetragonal structure was identified.
