In vitro comparison of physical characteristics of milled versus printed zirconia discs

Abstract

Purpose:

The purpose of this study was to compare the dimensional accuracy, translucency, and biaxial flexural strength of milled zirconia (MZ) versus 3D-printed zirconia (PZ) discs.

Materials & Methods:

A circular disc measuring 14.0 mm in diameter and 1.20 mm in thickness was designed using computer-assisted design (CAD) software. The resulting standard tessellation language (STL) file was used both as a control and to fabricate 36 zirconia (3Y-TZP) disc specimens (n = 36): 18 were milled (group MZ) and 18 were 3D-printed (group PZ). The diameter and thickness of each disc were measured using a digital caliper. Translucency was evaluated using a calibrated dental colorimeter. The flexural strength was determined using the piston-on-three-ball biaxial flexure test. All measurements were done by one blinded examiner. The statistical significance level was set to α = 0.05.

Results:

The MZ discs had significantly more accurate dimensions than the PZ discs in both diameter and thickness when compared to the control CAD software-designed disc. The MZ discs exhibited significantly higher translucency (translucency parameter (TP) = 16.95 ±0.36 vs. 9.24 ±1.98) and biaxial flexural strength (996.16 ±137.37 MPa vs. 845.75 ±266.16 MPa) than the PZ discs. Finally, MZ possessed a significantly higher Weibull modulus relative to PZ.

Conclusions:

The results showed that the milled specimens achieved better dimensional accuracy and were more translucent, stronger, and less prone to failure than printed specimens.

The use of zirconia ceramics has been revolutionary in restorative dentistry because of their superior biocompatibility, flexural strength, esthetics, and reduced preparation parameters.^1,2 The technology of computer-aided design and computer-aided manufacturing (CAD-CAM) is processed by two methods: milling or subtractive manufacturing (SM) and 3D-printing or additive manufacturing (AM).³

The most commonly used manufacturing method for zirconia ceramics is SM wherein a solid pre-sintered disc is modified using drills and sintered into the final designed restorative form.⁴ Several clinical studies have shown that SM produces clinically acceptable restorations.^5,6 However, SM does have some disadvantages. The contour of the milled restoration is limited by the size of the cutting drills and types of milling machines, and produces wasted debris that cannot be re-used.^3,7 Additionally, ‘tooling stress’ of damaged burs has been shown to cause micro fractures and crack propagation over time under occlusal loading.⁷

Several AM printing technologies exist include stereolithography, selective laser sintering, and fused deposition modeling. The AM technique can remove some of the limitations of SM, allowing for higher material utilization by appositional layering, the ability to produce more intricate designs, and no burs to wear.^2,7,8 The scope of 3D printing has covered the engineering of small, customized parts to personalized biomaterials and prosthetics including a kidney, leg, teeth, and blood vessels.⁹ The level of detail that 3D printing offers as new materials are developed and with the expiration of patents is very promising for the future of prosthodontics.⁹

A PubMed search using the keywords “3D printed” or “additive manufactured zirconia flexural strength” resulted in a total of only 34 articles in the ten-year period from 2013–2023. The reported flexural strength values for 3D-printed 3Y-TZP varied from 300 MPa to 1500 MPa, with a Weibull modulus ranging between m = 6 – 13.^8,10,11,12 Such broad variations in flexural strength and Weibull distribution suggest larger defects in 3D-printed zirconia relative to those of conventionally powder-pressed and milled zirconia. A number of other factors may have influenced the varied flexural strength, including thermocycling and chewing simulation,⁸ methodologies of measuring flexural strength, different dimensions of the test specimen, ^10,11 and different manufacturers.^10,12 In addition, there is limited research comparing the dimensional stability and translucency of AM versus SM zirconia. Therefore, the purpose of this research was to compare the dimensional accuracy, composition, density, translucency, and flexural strength of milled versus 3D-printed zirconia discs.

The null hypothesis is that there is no difference in diameter, thickness, translucency, and flexural strength of milled versus 3D-printed zirconia.

MATERIALS AND METHODS

Specimen preparation

A cylindrical disc measuring 14.0 mm in diameter and 1.20 mm in thickness was designed using computer-aided design (CAD) software (Blender, Amsterdam, Netherlands). The resulting standard tessellation language (STL) file (Fig 1) was used to fabricate 36 zirconia disc specimens (n = 36, divided into two groups of 18 each). Eighteen specimens were milled (group MZ) from an IPS e.max ZirCAD 3Y pre-sintered zirconia disc (3 mol% yttrium stabilized tetragonal zirconia polycrystals, 3Y-TZP, Ivoclar Vivadent, Schaan, Liechtenstein) using a five-axis milling machine (model TR5, vhf Inc., Hauppauge, NY). A 1 mm diameter diamond-coated triple tooth radius carbide cutter (Z100-R3D-40, vhf Inc.) was used to mill the specimens. The specimens were air-dried and brush-cleaned. Sintering was completed in a sintering oven (model baSiC Austromat 674, Dekema Dental, Freilassing, Germany). Specimens were gradually heated up to 1500° C, held at this temperature for 120 minutes, and then slowly cooled. The sintering process took approximately 10 hours.

FIGURE 1.

Screenshot depicting 14.00 mm x 1.20 mm disc designed using CAD software.

The other 18 specimens were 3D-printed (group PZ) with a 3 mol% TZP (LithaCon 3Y 230, Lithoz America, LLC, Troy, NY) using a lithography-based ceramic manufacturing (LCM) process (Lithoz Cerafab 7500 Dental 3D-printer). Each layer was printed with a layer thickness of 10 microns, at a speed of 100 layers per hour following all manufacturers’ recommendations. The specimens were built at a 0° orientation, i.e., the layers were stacked in the direction along the central axis of the circular disc. A previous study on 3D-printed Zpex Smile zirconia powder (Tosoh Corp., Tokyo, Japan) using the same LCM technology has shown that flexural strength was the highest for specimens with 0° build orientation.¹³ Each specimen was cleaned with a cleaning fluid (LithaSol, Lithoz America, LLC, Troy, NY) and then thermally processed to dry at 120° C. Debinding of ceramic-filled photopolymers was completed at 1000° C and final sintering to 1450° C was achieved with a 2 hour dwell time.

Dimensional accuracy

The diameter and thickness of each completed zirconia disc specimen were measured using a calibrated digital caliper (Mitutoyo, Absolute 500–196-20, Kanagawa, Japan). The diameter of each specimen was measured 3 times by the same blinded operator to minimize the possible operator error.

Composition, density, and translucency

The crystalline phases in sintered zirconia specimens were analyzed by X-ray diffraction (XRD) with CuKα radiation (Rigaku Miniflex 6G, Tokyo, Japan). Specimens (n = 6) were mounted on the specimen stage with the top side against the flat surface of a stub. Spectra were collected over a 2θ range of 20° – 80° with a step size of 0.01° and a scan rate of 2° per min.¹⁴ The tetragonal and cubic phases were calculated based on the peak intensities in the 2θ range between 72° – 76°. The (004)_t, (220)_t, and (400)_c peaks were used to measure their intensities and calculate both tetragonal and cubic phase fractions according to the equation as follows:

$$X_c=\frac{I_c\left(400\right)}{I_c\left(400\right)+I_t\left(004\right)+I_t\left(220\right)}$$ (Eq.1)

$$X_t=\frac{I_t\left(004\right)+I_t\left(220\right)}{I_c\left(400\right)+I_t\left(004\right)+I_t\left(220\right)}or1-X_c$$

where $X_c$ and $X_t$ represent the proportions of cubic and tetragonal phases respectively. The $I_t$ and $I_c$ represent the intensities of the corresponding peaks. ^15,16

The bulk density of the zirconia discs ($n=10$) was measured using the Archimedes water displacement method described by Tong et al.¹⁷ Relative density (RD) of the two test groups were calculated using the experimentally determined bulk density values divided by the theoretical density value (6.10 g/cm³) for 3Y-TZP.¹⁸

Specimen translucency parameters (TP) ($n=10$) were measured using a dental colorimeter (SpectroShade Micro, MHT, Niederhasli, Switzerland). For objects commonly viewed in reflection (such as dental restorations), the TP is the most used and standardized method.¹⁹ Briefly, color coordinates CIE $L*a*b*$²⁰ were measured over standard backgrounds (black $L*=1.8$, $a*=1.3$, $b*=-1.5$ and white $L*=95.7$, $a*=-1.3$, $b*=2.6$). For optical continuity, a drop of coupling liquid (refractive index: 1.8, Gem Refractometer Liquid, Cargille Laboratories, Inc., Cedar Grove, NJ, USA) was placed between the background and the specimen.²¹ The TP values ($n=10$) were determined by the color difference between the specimen on black ($B$) and white (W) backgrounds, according to Equation 2:

$$\text{TP}=\sqrt{({\text{L}}_{\text{B}}^{*}-{\text{L}}_{\text{W}}^{*}{)}^2+({\text{a}}_{\text{B}}^{*}-{\text{a}}_{\text{W}}^{*}{)}^2+({\text{b}}_{\text{B}}^{*}-{\text{b}}_{\text{W}}^{*}{)}^2}$$ (Eq.2)

where $L*$, $a*$, and $b*$ refer respectively to the lightness, redness to greenness, and yellowness to blueness coordinates in the CIE color space.²²

Flexural strength

A biaxial flexural strength piston-on-three-ball test (Fig 2) was used to obtain the fracture load of each specimen in Newtons using a dual column tabletop model universal testing machine (model l 68TM-5, Instron, Norwood, MA). Both the loading piston and support balls were made from hardened steel. The diameter of the loading piston was 1.40 mm and the diameter of the support ball was 3.20 mm, following International Organization for Standardization ISO/ FDIS 6872:2014(E)²³ (Fig 3). Teflon tape was placed between the support balls and the specimens to reduce the friction while scotch tape was placed between the load piston and the specimens to prevent contact damage. All specimens were subjected to a loading rate (crosshead speed) of 1 mm/min until fractured. The biaxial flexural strength of each specimen was then calculated in Megapascals (MPa) using Equation 3, following ISO/FDIS 6872:2014(E).

$$\sigma =\frac{-\mathrm{0,2387}P(X-Y)}{b^2}$$ (Eq.3)

where

$$X=(1+v)\ln {(r_2/r_3)}^2+[(1-v)/2]{(r_2/r_3)}^2$$

$$Y=(1+v)[1+\ln {(r_1/r_3)}^2]+(1-v){(r_1/r_3)}^2$$

FIGURE 2.

Piston-on-three-ball biaxial flexural strength test.

FIGURE 3:

Piston-on-three-ball biaxial strength test parameters. ¹⁰

Where $r_1=5.00$ mm (radius of support circle), $r_2=0.70$ mm (radius of the loaded area), $r_3=7.00$ mm (disc radius), $b=1.20$ mm (disc thickness), $v=3.15$ (Poisson’s ratio for 3Y-TZP), and $P$ = the total load causing fracture (N).

Weibull analysis

In order to predict the fracture resistance of these 3Y-TZP ceramics, the Weibull failure probabilities were also determined. The Weibull failure probability is described by the Weibull modulus $m$; a higher $m$ value suggests a smaller scatter in measured zirconia strength. For critical flexural stress (strength) $s_F$ of zirconia discs, the Weibull failure probability P can be defined as:

$$P=1-\text{exp}[-{({\sigma }_F/{\sigma }_0)}^m]$$ (Eq.4)

where ${\sigma }_0$ is a scaling stress or the short-term biaxial 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 lnσ_F gives a straight line with slope $m$:

$$ln(ln(1/(1–P)))=m(ln{\sigma }_F–ln{\sigma }_0)$$ (Eq.5)

In a practical sense, Weibull curves reported in this study were obtained by fitting the scatter plot of ‘cumulative probability versus log (measured strength data)’ using the least-square method of linear regression with a 95% confidence interval in SigmaPlot Version 12 (Inpixon, Palo Alto, CA). The slope of the line of best fit represents the Weibull modulus, $m$.

Statistical analysis

The diameter and thickness of each specimen (n = 18) were compared to the original STL dimensions of 14.0 mm diameter and 1.20 mm thickness serving as a control, to assess dimensional accuracy. Specimen data were summarized as means and standard deviations (SDs). An independent samples t-test was used to compare mean values between groups, unless within-group variances were heterogeneous, in which case the Welch t-test was substituted. Cohen’s d was used to index the effect size.

Specimen size was set at 18 specimens per group in order to detect a group difference of at least 1 SD in a 2-tailed t-test with type 1 and 2 error rates of 5 and 20%, respectively.

The obtained Weibull moduli $m$ were statistically compared within the group as well as between the groups by critically examining the difference between the slopes from two independent samples using both pooled and unpooled error variances according to Howell.²⁴

The p-values for each statistical test are shown in the results section below and the statistical significance level was set to α = 0.05.

RESULTS

Dimensional accuracy

Group PZ had an average diameter of 14.04 ±0.06 mm and group MZ had an average diameter of 14.01 ±0.03 mm (Welch test, p = 0.225, d = 0.42). While the means were similar, the standard deviation of group PZ was approximately twice that of group MZ. This reflects the larger range seen in diameter for group PZ (13.92 – 14.14 mm) than for group MZ (13.97 – 14.09 mm) (Fig 4).

FIGURE 4:

Box chart showing less accurate and more variable diameter of PZ specimens.

Group PZ had an average thickness of 1.22 ±0.05 mm, significantly larger than that of group MZ, which had an average thickness of 1.19 ±0.01 mm (Welch test, p = 0.049, d = 0.70). The standard deviation of group PZ was five times that of group MZ. This again reflects the greater range in the thickness for group PZ specimens (1.14 – 1.31 mm) than for group MZ specimens (1.18 – 1.23 mm) (Fig 5).

FIGURE 5:

Box chart showing the lower accuracy and less consistent thickness of PZ specimens. The red dot represents an outlier data points further than 1.5 times the interquartile range (IQR).

Because specimens could be either larger or smaller than designed, groups were also compared on their absolute deviation from each design parameter. Analysis showed significantly larger average absolute deviations in diameter in group PZ, 57.22 ±42.3 μm, than in group MZ, 25.6 ±22.0 μm (p = 0.009, d = 0.94). Analysis also showed significantly larger average absolute deviations in thickness in group PZ, 47.8 ±30.2 μm, than in group MZ, 13.3 ±8.44 μm (p < 0.001, d = 1.55).

Composition, density, and translucency

The phase assembly, bulk density, and TP of groups PZ and MZ are summarized in Table 1. Since the translucency of the specimens’ changes with the different thicknesses,²⁵ both milled and printed zirconia discs were 1.20 mm thick to enable direct comparisons of TP values of both manufacturing methods. Both groups contained approximately 83 wt.% tetragonal zirconia phase and ~17 wt.% cubic phase. This is because both the LithaCon 3Y 230 (PZ) and IPS e.max ZirCAD (MZ) materials were fabricated from the same Zpex grade zirconia raw powder (Tosoh Corp. Tokyo, Japan).

Table 1:

Overview of the study results comparing the properties of CAD-CAM milled zirconia (MZ) versus 3D-printed zirconia (PZ). Values are presented in mean (standard deviation).

	MZ	PZ
Phase fractions (%)
t-ZrO2	82.69 (0.10)	82.77 (0.29)
c-ZrO2	17.31 (0.10)	17.23 (0.29)
Bulk density, ρ (gm/cm2)	6.08 (0.01)	6.06 (0.02)
Relative density, RD (%)	99.67	99.34
Translucency parameter, TP (%)	16.95 (0.36)	9.24 (1.98)
Biaxial flexural strength, σ (MPa)	996.16 (137.37)	845.75 (266.16)
Weibull modulus, m	7.27	7.23/1.92

Flexural Strength

Post-mortem optical microscopic analysis revealed that all specimens fractured from the tensile surface that faced the support balls. Fracture origins were all located near the center of the disc, directly beneath the load piston. None of the fractures occurred at the support balls.

The mean biaxial flexural strength of group PZ was 845.75 ±266.16 MPa while that of group MZ was 996.16 ±137.37 MPa. Thus, the biaxial flexural strength of group PZ was about 150.39 MPa lower than group MZ (Welch test p = 0.043, d = 0.71). For the biaxial flexural strength measurements, group PZ had an SD approximately double that of group MZ, resulting in more consistent measurements of biaxial flexural strength for group MZ (707.49 – 1190.00 MPa) than group PZ (326.71 – 1134.78 MPa). (Fig 6) Importantly, group PZ had some specimens of notably lower strength. For example, specimen P4 = 326.71 MPa, and specimen P9 = 370.49 MPa (Table 2). Four specimens in group PZ had a biaxial flexural strength below 707 MPa, while no specimens in group MZ were lower than 707 MPa (Table 3). Similarly, only 6 of 18 (33.33%) of group PZ specimens had a biaxial flexural strength above 1,000 MPa, while 9 out of 18 (50.00%) of group MZ specimens were above 1,000 MPa.

FIGURE 6:

Box chart showing the higher and more consistent biaxial flexural strength of the MZ specimens.

Table 2:

Diameter (d), thickness (b) and strength (σ) of PZ (printed zirconia) specimens.

Specimen	d (mm)	b (mm)	Strength σ (MPa)
P1	13.92	1.30	423.47
P2	14.01	1.23	915.26
P3	14.01	1.23	964.11
P4	14.03	1.29	326.71
P5	13.98	1.24	993.50
P6	14.06	1.31	525.00
P7	13.96	1.27	918.68
P8	13.95	1.24	1027.48
P9	14.05	1.28	370.49
P10	14.06	1.17	1118.81
P11	14.11	1.18	946.90
P12	14.14	1.17	799.32
P13	13.99	1.18	1134.78
P14	14.03	1.19	1070.08
P15	14.02	1.18	1085.84
P16	14.07	1.15	809.38
P17	14.10	1.17	1075.51
P18	14.14	1.14	718.19

Table 3:

Diameter (d), thickness (b) and strength (σ) of MZ (milled zirconia) specimens.

Specimen	d (mm)	b (mm)	Strength σ (MPa)
M1	14.03	1.18	1052.63
M2	13.97	1.18	1101.52
M3	14.03	1.19	906.38
M4	14.01	1.18	1061.43
M5	13.99	1.19	981.65
M6	13.99	1.18	1190.00
M7	14.00	1.19	1151.95
M8	13.98	1.19	921.91
M9	14.01	1.19	1033.12
M10	13.99	1.18	994.88
M11	14.05	1.18	1022.89
M12	14.03	1.19	1096.84
M13	14.09	1.23	707.49
M14	13.98	1.20	988.46
M15	14.05	1.20	765.81
M16	14.02	1.20	794.34
M17	14.04	1.19	972.02
M18	14.00	1.18	1187.62

Weibull modulus

Weibull plots for biaxial flexural strength data of MZ and PZ are shown in Figure 7. The Weibull strength distribution of group MZ represented a typical failure trend for brittle materials. A Weibull modulus m = 7.27 along with tight 95% confidence bounds is consistent with literature values for conventionally fabricated 3Y-TZP.²⁶

FIGURE 7:

Weibull plots for biaxial flexural strength data of MZ and PZ.

The Weibull distribution for biaxial flexural strength of group PZ exhibited two distinct regions. For the upper 50% strength specimens, the Weibull modulus was m = 7.23, almost identical to that (m = 7.27) of MZ (p > 0.1 for both pooled and unpooled error variances). But for the lower half strength PZ specimens, the Weibull modulus m = 1.92 was significantly lower than that for the upper 50% strength specimens as well as for group MZ (p < 0.001 for both pooled and unpooled error variances). The extremely low m value for the lower half strength of group PZ suggests that these specimens contain large strength-limiting flaws.

DISCUSSION

The present study evaluated the dimensional accuracy, density, translucency parameter, and biaxial flexural strength of milled versus 3D-printed 3Y-TZP ceramics. The null hypothesis that there were no differences between the two groups in these parameters was rejected.

While the mean diameters in the two groups appeared similar, a result of positive and negative deviations canceling each other, absolute deviations from the design parameter were significantly larger in group PZ. Mean thickness was also significantly greater, and further from the design parameter, in group PZ. In addition, group PZ specimens were more variable than group MZ specimens, as reflected by higher standard deviations and larger ranges of obtained values. Thus, milling produced more accurate and precise results than printing.

The wide variations in the dimensions of the group PZ may be due to the use of support struts. After the completion of the printing and curing process, these support struts are manually removed and polished which can introduce errors in dimensions. On the contrary, greater dimensional stability, and higher translucency and biaxial flexural strength were found in the specimens of group MZ. This is consistent with several previously published studies.^1,2,11,27,28 One study theorized that support structures present in AM (but not present in SM) may contribute to the lower dimensional stability of group PZ.² Upon the removal of the support structures, material may be removed and then polished. Zenthofer et al demonstrated up to 200 microns of material may be removed upon polishing.²⁷ Lerner et al reported this will result in dimensional change,² so no finishing or polishing of the specimens was done, as has been reported previously.^11,12 Therefore, the locations of the support struts should not be placed in critical areas where the accuracy of occlusion, interproximal contacts, and marginal fit are impacted.²

Despite the near identical phase compositions between the two groups, group MZ possessed a higher bulk density and a much higher TP value relative to its PZ counterparts. Previous studies showed that a small amount of porosity (up to 1 – 3 vol.% or so) may not have profound detrimental effects on the strength of 3Y-TZP, provided the pore sizes are not larger than those strength-limiting flaws in the microstructure.²⁹ However, porosities even as low as 0.05% and pore sizes as small as 100 nm could have a deleterious effect on the translucency of 3Y-TZP.³⁰ This is because the large difference in refractive index between zirconia (n ~ 2.21) and air (n ~ 1.00) results in significant light scattering at pores, thus reducing translucency.

Regarding biaxial flexural strength, milling produced stronger and more consistent specimens than printing. However, there was a much wider variation of measurements in the specimens among group PZ. According to the ISO/FDIS 6872, the minimum required flexural strength for a monolithic or fully covered (layered) substructure for a three-unit prosthesis including a molar is 500 MPa.²³ Weibull failure prediction using Equation 5 showed that for group PZ, 16.3% would not meet this 500 MPa threshold. In fact, 5% of failures would occur at 308.41 MPa and 10% of failures at 408.10 MPa. For group MZ, only 0.5% of specimens would not meet the 500 MPa threshold. 5% of failures would occur at 686.08 MPa and 10% of failures at 757.18 MPa.

There is speculation that the reason for the greater biaxial flexural strength of milled zirconia is because of the manufacturing process of pre-sintered zirconia. Fabrication under isostatic pressure results in greater density and fewer porosities.¹ On the other hand, AM requires the depositions of multiple layers without pressure during printing and requires the binder to be burned out (debinding) prior to sintering which may result in more voids and defects. These defects may limit its strength.^{1, 27} Better printing techniques are being developed. Lithoz now has a furnace that allows debinding and sintering of zirconia in a single step which may reduce porosities.³¹

In this study, cylindrical discs were chosen for test specimens and a biaxial flexural strength test was used to assess flexural strength because bars used as specimens often have processing flaws along their edges.³² When strength measurements are done using 3-point and 4-point bending tests, fracture often originates from the edge of the specimen. As a result, it does not reflect the intrinsic strength of the material.^23,33 One study showed inconsistent measurement of data when the 3-point bending test was used.³⁴ The use of a biaxial flexural strength test produces more accurate measurements since edges are not subject to loading.³²

Clinicians have many materials to choose from when providing dental restorations. Each material has advantages, disadvantages, indications, and contraindications. It is important for the success and survival of prostheses that clinicians understand and know the predictability, reproducibility, strength, and translucency of the material being used. These results showed an increased risk of strength-limited failures in the printed specimens and favored the consistently strong and translucent specimens produced by milling. Nevertheless, these results must be weighed with other studies, showing higher flexural strength for the AM zirconia.¹²

Limitations of this study must be considered. A limitation of this study used discs, rather than anatomically contoured specimens. Future studies should compare the physical characteristics of zirconia with 2 different manufacturing techniques and anatomic contours. Only one brand of zirconia (raw powder) was used for each manufacturing method and no aging or thermocycling was done. Two previous studies, however, showed no difference in flexural strength between PZ and MZ after thermocycling.^26,35 The orientation angle of printing was not examined. One study using x-ray tomography showed differences in horizontally versus vertically printed specimens.³⁶ Lastly, in this study only one resolution of printing was examined. Perhaps printing at different resolutions and orientations may yield different results.

CONCLUSIONS

Milled zirconia specimens were shown to be more accurate and precise in achieving a designed diameter and thickness and were more dense, translucent, and stronger than printed discs.