The effect of surface grinding and sandblasting on flexural strength andreliability of Y-TZP zirconia ceramic

T. Kosmaca,* , C. Oblakb, P. Jevnikarb, N. Fundukb, L. Marionb

aCeramics Department, Jozef Stefan Institute, Jamova 39, 1001 Ljubljana, SloveniabDepartment of Prosthodontics, Medical Faculty, University of Ljubljana, Ljubljana, Slovenia

Received 1 January 1999; received in revised form 8 June 1999; accepted 23 June 1999

Abstract

Objectives:This study was conducted to evaluate the effect of grinding and sandblasting on the microstructure, biaxial flexural strengthand reliability of two yttria stabilized tetragonal zirconia (Y-TZP) ceramics.

Methods:Two Y-TZP powders were used to produce fine grained and coarse grained microstructures. Sixty discs from each material wererandomly divided into six groups of ten. For each group, a different surface treatment was applied: dry grinding, wet grinding, sandblasting,dry grinding1 sandblasting; sandblasting1 dry grinding and a control group. Biaxial flexural strength was determined and data wereanalyzed using one-way ANOVA, followed by Tukey’s HSD test�p , 0:05�. In addition, Weibull statistics was used to analyze thevariability of flexural strength. The relative amount of transformed monoclinic zirconia, corresponding transformed zone depth (TZD)and the mean critical defect sizeccr were calculated.

Results:There was no difference in mean strength between the as sintered fine and coarse grained Y-TZP. Significant differences�p ,0:05� were found between the control group and ground fine grained material for both wet and dry grinding. Sandblasting significantlyincreased the strength in fine and coarse grained materials. All surface treatment procedures reduced the Weibull modulus of Y-TZP. Forboth materials, the highest amount of the monoclinic phase and the largest TZD was found after sandblasting. Lower amounts of themonoclinic phase were obtained after both grinding procedures, where the highest mean critical defect sizeccr was also calculated.

Significance:Our results indicate that sandblasting may provide a powerful technique for strengthening Y-TZP in clinical practice. Incontrast, grinding may lead to substantial strength degradation and reduced reliability of prefabricated zirconia elements, therefore, sand-blasting of ground surfaces is suggested.q 1999 Academy of Dental Materials. Published by Elsevier Science Ltd. All rights reserved.

Keywords:Zirconia ceramic; Flexural strength; Reliability; X-ray diffraction

1. Introduction

Yttria stabilized tetragonal zirconia (Y-TZP) is gainingrecognition as a candidate material in dentistry owing to itsgood mechanical properties. It is currently used as a corematerial in full-ceramics dental restorations [1–3], implantsuperstructures [4] and orthodontic brackets [5]. Comparedto other dental ceramics, superior mechanical properties aredue to the transformation toughening mechanism, similar tothat exploited in quenched steel. ZrO2 is a polymorphicmaterial that has three allotropes: the monoclinic phase isstable up to 11708C where it transforms into the tetragonalphase, which is stable up to 23708C, and the cubic phaseexists up to the melting point at 26808C [6]. Of greatestsignificance for the mechanical properties is the tetragonal

(t)! monoclinic (m) phase transformation that occurs by adiffusionless shear process at near sonic velocities, similarto those of martensite formation in quenched steel [7]. Rela-tively large volume expansion (3–5%) associated with thet ! m phase transformation, which can be induced by anexternally applied stress exerted by grinding, impact or frac-ture, leads to the development of internal stresses opposingthe opening of the crack, and therefore acting to increase theresistance of the material to crack propagation [8]. This isthe reason why Y-TZP and other zirconia toughened cera-mics exhibit relatively high fracture toughness compared toconventional brittle ceramics, which in turn may also resultin high strength. There is a linear relationship betweenstrength and toughness for a brittle material [9]

df � 1w

KIC����ccrp �1�

whered f is the fracture stress,w the geometric constant,KIC

Dental Materials 15 (1999) 426–433

dentalmaterials

0109-5641/99/$20.00 + 0.00q 1999 Academy of Dental Materials. Published by Elsevier Science Ltd. All rights reserved.PII: S0109-5641(99)00070-6

www.elsevier.com/locate/dental

* Corresponding author. Tel.:1 386-61-177-3227; fax:1 386-61-219-385.

E-mail address:Tomaz.Kosmac@ijs.si (T. Kosmacˇ)

the fracture toughness andccr the mean critical defect size toinitiate failure. The value ofccr is determined by the inherentflaw population introduced during material processing andcomponent fabrication [10]. Thus, zirconia ceramic willexhibit high strength provided that the critical defect sizeis kept small, which necessitates optimization of the cera-mics manufacturing process. Furthermore, grinding with itsability to transform zirconia in the surface region, can beutilized to introduce residual surface compressive stressesthat increase the mean flexural strength of zirconia tough-ened ceramics considerably [11–13]. The extent of strength-ening depends on the vol% of transformed ZrO2 e.g. on thedepth of the surface compressive layer, which increaseswith the thermodynamic instability of the tetragonal ZrO2

grains and grinding severity [12]. On the other hand, severegrinding introduces deep surface flaws which act as thestress concentrators, and may become strength determiningif their length largely exceeds the depth of the grindinginduced surface compressive layer [14]. In this case, theoverriding grinding factor in determining the flaw sizewill be the size of the abrasive grains; the coarser thegrain size the larger will be the flaw size [15].

The reported manufacturing technology of zirconia postand core systems, implant superstructures and orthodonticbrackets involves shaping and sintering of ceramic pre-forms with subsequent machining to adjust a final formand sandblasting to improve adhesion during cementation[1,16]. It is, therefore, likely that the surface of the meta-stable tetragonal zirconia ceramic will be transformed, i.e.constrained, and also damaged, which will influence themechanical properties and reliability of the material.

The purpose of our study was, therefore, to evaluate theeffect of grinding and sandblasting on the biaxial flexuralstrength and reliability of yttria stabilized tetragonal zirco-nia (Y-TZP) ceramic containing 3 mol% Y2O3. Since the

transformability and thereby related fracture toughness ofTZP ceramics depend not only on the chemical composi-tion, but also on the grain size [17], the aim of our work wasalso to compare two Y-TZP ceramics of the same chemicalcomposition, differing in mean grain size.

2. Materials and methods

Two commercially available high purity Y-TZP powderswere used to produce specimens: a low calcined TZ-3YBgrade (Lot No. Z 305235B, Tosoh, Tokyo, Japan) with ahigher specific surface area (16 m2/g) and a high calcinedTZ-3YSB grade (Lot No. S 306324B), with lower specificsurface area (7 m2/g). Both powders were of the samenominal chemical composition, containing 3 mol% Y2O3

in the solid solution. Due to its higher sinter-activity, theTZ-3YB powder compacts densify at lower temperaturesresulting in finer microstructure, whereas higher sinteringtemperatures are needed to reach a nearly theoretical densityof the TZ-3YSB powder compacts, resulting in coarsermicrostructure. Both powder grades were ready-to-pressgranulates containing a few percent of an acrylic binder.A series of TZ-3YB (fine grained) and TZ-3YSB (coarsegrained) pellets (20 mm in diameter and 2 mm thick) wereuniaxially dry pressed at 15 MPa and subsequently sinteredin air at 15008C for 2 h and at 15508C for 4 h, respectively.The relative density of sintered specimens was determinedby the Archimedes method using distilled water as immer-sion liquid. Grain size determinations were made on SEMmicrographs of polished and thermally etched (14508C,30 min) specimens, using semiautomatic image analysis tocalculate equivalent spherical diameters.

After firing, 60 disc-shaped specimens (15:5^ 0:03 mmin diameter and 1:5^ 0:03 mm thick) from each materialwere randomly divided into six groups of 10. For eachgroup, a different surface treatment was applied to the topsurface of the specimens, as indicated in Table 1. A coarsegrit (150mm) diamond burr (ISO-No. 806 314 110 534 018,Dica, Dendia, Austria), mounted on a high-speed handpiece,was chosen for surface grinding in order to simulate clinicalconditions. For sandblasting, discs were mounted in aspecial sample holder at a distance of 30 mm from the tipof the sandblaster unit (Duostar Z2, Bego, Bremen,Germany), equipped with a nozzle of 5 mm in diameter.The combination of two zirconia powders and six surfacetreatments yielded 120 specimens. After surface treatment,the specimens were cleaned for 3 min in an ultrasonic bathcontaining ethylacetate. Dimensions of the specimens weremeasured with a digital caliper (Digimatic CD-4 BS, Mitu-toyo Corporation, Japan). The thickness of each specimenwas calculated as the mean of three measurements taken atrandom sites.

Before and after the surface treatment, the samples wereverified by X-ray diffractometry (XRD, D 5000, Siemens,Germany), using CuKa radiation. The relative amount of

T. Kosmacˇ et al. / Dental Materials 15 (1999) 426–433 427

Table 1Surface treatment procedures for fine and coarse grained Y-TZP zirconiaceramic

Surface treatment proceduresAs sintered Control groupsDry ground Grinding for 30 s with coarse grit (150mm)

diamond burr without water spray coolingWet ground Grinding for 30 s with coarse grit (150mm)

diamond burr with water spray coolingSandblasted Sandblasting for 15 s with Al2O3 particles

(average particle size� 110mm) at pressureof 4 bars

Dry ground1 sandblasted Grinding for 30 s with coarse grit (150mm)diamond burr without water spray coolingfollowed by sandblasting for 15 s with Al2O3

particles (average particle size� 110mm) atpressure of 4 bars

Sandblasted1 dry ground Sandblasting for 15 s with Al2O3 particles(average particle size� 110mm) at pressureof 4 bars followed by grinding for 30 s withcoarse grit (150mm) diamond burr withoutwater spray cooling

transformed monoclinic zirconia on the treated surfaces wasdetermined from the integral intensities of the monoclinicM(111) and M(111), and the tetragonal T(111) peaksaccording to the method of Garvie and Nicholson [18],which is most commonly applied to determine the phasecomposition of zirconia powders and compacts withrandomly distributedm and t phase at any distance fromthe surface exposed to XRD analysis:

XM �IM�111� 1 IM�11�1�

IM�111� 1 IM�11�1� 1 IT�111��2�

The thickness of the transformed surface layer of surfacetreated samples was calculated using the X-ray determina-tion method originally proposed by Kosmac et al. [19].According to this method, the transformed zone depth(TZD) can be calculated from the relative amounts of themonoclinic phase, assuming that, within the transformedsurface layer, all the tetragonal grains have transformedinto the monoclinic symmetry:

TZD � sinu

2mln

11 2 XM

�3�

whereu � 158 is the angle of reflection,m � 0:0642 theabsorption coefficient andXM the relative monoclinic frac-tion obtained from the XRD analysis on the basis of Eq. (2).Although Eq. (3) yields conservative values for the TZD, itcan be used to compare the influence of various surfacetreatments on the thickness of the surface compressivelayer for the two Y-TZP ceramics.

The indentation technique was used to determine fracturetoughness of fine and coarse Y-TZP. A load of 200 N wasapplied to the specimens with a Vickers hardness indenter.Five specimens from each material were used and threeindentations were made on each sample. Fracture toughnesswas calculated using the length of the cracks emanatingfrom the Vickers impression. The observations wereanalyzed according to the expression for the stress intensity

factor,KIC, given by Chantikul et al. [20]:

KIC � 0:016�E=H�1=2P=c3=2 �4�whereE is Young’s Modulus,H the Vickers hardness,P theindentation load andc the crack length as measured from thecenter of impression. Statistical significance of differencesbetween both materials was analyzed with unpairedt-test.

Prior to strength testing on a universal testing machine(Model 4301, Instron Corp., Canton, USA), all specimenswere stored for 24 h in distilled water at 378C. Test speci-mens were fractured in biaxial flexure by means of thepiston-on-3-ball technique, with the surface treated sideunder tension, at a loading rate of 1 mm/min. The discswere concentrically supported on three ball bearings�diameter� 1:6 mm� that were evenly spaced around asupport circle having a radius of 11 mm. The specimenswere loaded to fracture by a compressive load applied bya circular cylinder with a diameter of 1.6 mm. All measure-ments were made in a thermostatically-controlled waterbath kept at 378C. The load to failure was recorded foreach disc and flexural strength was calculated using theequations for biaxial flexure strength, developed by Wacht-man et al. [21]. The statistical significance of differencesbetween data were analyzed with ANOVA, followed byTukey’s HSD test at a significance level of 0.05. The varia-bility of the flexural strength values was analyzed using theWeibull distribution function:

P�s� � 1 2 exp�2s=s0�m �5�whereP is the fracture probability,s the fracture strength,s0 the characteristic strength at the fracture probability of63.2% andm is the Weibull modulus which is the slopeof the ln�ln 1=1 2 P� vs. lns plots. The statistical variabilityof estimates for the two characteristic Weibull parametersmands0 were calculated according to Ritter et al. [22]. Theseauthors have demonstrated that Weibull modulusmdependsonly upon the sample size, i.e. upon the number of speci-mens per condition, whereas the variability in the character-istic strengths0 depends upon both sample size andm.

T. Kosmacˇ et al. / Dental Materials 15 (1999) 426–433428

Fig. 1. SEM micrographs showing microstructures of polished and thermally etched (a) fine and (b) coarse grained Y-TZP ceramics after sintering.

3. Results

After sintering, the fractional density of fine grained (TZ-3YB) and coarse grained (TZ-3YSB) zirconia exceeded99% of the theoretical value. The SEM micrographs ofsintered, polished and thermally etched Y-TZP samplesare shown in Fig. 1, illustrating dense, uniform microstruc-tures. The quantitative microstructural analysis revealednearly log-normal grain size distribution with the meangrain size of 0.31 and 0.44mm for the fine and coarsegrained Y-TZP, respectively.

According to the XRD pattern of as sintered Y-TZP cera-mics, both the fine grained and the coarse grained materialconsisted of 100% tetragonal zirconia. After surface treat-ment, however, detectable monoclinic peaks with a markedpreference of the M(111) orientations appeared in the XRDpattern, as shown in Fig. 2 for the coarse grained material. Asimilar XRD pattern was obtained with fine grained material.

The relative amounts of the monoclinic zirconia onsurface treated samples are listed in Table 2 together withthe corresponding TZD values. For both materials, the high-est amount of the monoclinic phase was found after (i)sandblasting and (ii) dry grinding followed by sandblasting.Lower amounts of them phase were obtained after bothgrinding procedures. When sandblasting was followed bydry grinding the amount of the monoclinic phase dropped ascompared to sandblasted group and was found to be wellwithin the values of both ground groups.

The TZD, as calculated from the relative amounts of themonoclinic phase follows the same sequence as theXM

values. The mean fracture toughness of as sintered finegrained Y-TZP was 4:45^ 0:05 MPa m1=2

; significantly�p , 0:05� less than in coarse grained material, which wasmeasured to be 4:76^ 0:11 MPa m1=2

:

The mean values of biaxial flexural strength in MPa andthe respective standard deviations are reported in Table 3.

T. Kosmacˇ et al. / Dental Materials 15 (1999) 426–433 429

Fig. 2. XRD pattern obtained from as sintered, dry ground and sandblasted surface of coarse grained Y-TZP ceramic.

Table 2Relative amounts of monoclinic zirconia (XM in %) verified by X-ray diffractometry and transformation zone depth (TZD) of surface treated Y-TZP ceramic

Surface treatment Relative amounts of monoclinic phase in % (SD) Transformation zone depth TZD inmm (SD)

Fine Y-TZP Coarse Y-TZP Fine Y-TZP Coarse Y-TZP

Dry ground 4.3 (0.6) 5.3 (0.6) 0.09 (0.012) 0.11 (0.013)Wet ground 3.4 (0.7) 4.2 (0.4) 0.07 (0.015) 0.09 (0.007)Sandblasted 13.9 (1.2) 15.2 (2.3) 0.30 (0.027) 0.33 (0.054)Dry ground1 sandblasted 12.7 (1.9) 15.7 (1.3) 0.27 (0.046) 0.34 (0.032)Sandblasted1 dry ground 3.7 (0.9) 3.1 (1.2) 0.05 (0.031) 0.06 (0.025)

There was no difference in mean strength between the assintered fine and coarse grained Y-TZP ceramic. Statisti-cally significant differences�p , 0:05� were found betweenthe control group and ground fine grained material for bothwet and dry grinding. There was, however, no differencebetween the control group and the two grinding proceduresin the coarse grained material. No difference was alsoobserved between dry and wet grinding in both materials.Sandblasting increased the strength, and there were statisti-cally significant differences between the control group (assintered) and sandblasting in fine and coarse grained mate-rials �p , 0:05�: When both surface treatment procedureswere applied in a sequence, the strength values were signif-icantly lower �p , 0:05� as compared to control groups,when samples were first sandblasted and than dry groundin both materials. Dry grinding followed by sandblastingalso revealed higher strength than the control groups,whereas significant differences were only found in thecoarse grained material.

Weibull statistical analysis of the biaxial flexural strengthdata yielded two characteristic parameters for each testgroup, the characteristic strength,s0 (which is the scaleparameter) and the Weibull modulus,m (which is theslope parameter). The results are summarized in Table 3.For both materials, highestm-value was obtained with thecontrol group, followed by the sandblasted group. If grind-ing was involved in the surface treatment (wet or dry, priorto- or after sandblasting), the Weibull modulus of finegrained material dropped to about one-half of its initialvalue (of the control group), and that of coarse grainedmaterial to about one-third of its initial value, them valuesof sandblasted group being in between.

4. Discussion

The results of the present investigation revealed thatsurface grinding and sandblasting exhibit a counteractingeffect on the strength of yttria stabilized tetragonal zirconia(Y-TZP) ceramic. Surface grinding using a coarse grit

(150mm) diamond burr at a high rotation speed loweredthe mean strength and reliability, whereas sandblastingimproved the mean strength, at the expense of lower relia-bility. During grinding, tens of microns of material wereremoved by a single pass as the burr was moved back andforth across the surface, and sparks were commonlyobserved, indicating that both stresses and temperatureswere high during this operation. Nevertheless, the amountof transformed zirconia on ground surfaces was low as wasthe calculated depth of the surface compressive layer contri-buting to the strengthening of the material. Previous workhas shown that severe machine grinding is less effective ininitiating the t ! m transformation in TZP materials [23–25]. In addition, Swain and Hannink [26] reported that handground Ce-TZP surface contained about five times moremonoclinic zirconia than severely machine ground surfaceof the same material. This, they argued, is because extensiveheat is generated during severe machine grinding in spite ofa stream of coolant that was directed near the cutting edgeduring grinding. As a result, locally developed temperaturesexceeded them! t transformation temperature and thereversem! t transformation occurred. In contrast, theforward t ! m transformation was retained upon handgrinding at lower speed and grinding force, which was notassociated with extensive heat generation. Therefore, basedon almost negligible amounts of transformed zirconia upongrinding under conditions used in our work, we assume thatthe locally developed temperatures exceeded them! ttransformation temperature (about 7008C) above whichthe tetragonal zirconia is thermodynamically stable. Onthe other hand, due to high stresses developed during grind-ing, severe surface cracks must have been formed whichlowered the strength and reliability of the material.

Sandblasting, in contrast, is considered to be a moregentle process, during which considerably less material isremoved from the surface. In spite of lower stresses occur-ring during sandblasting, the thickness of the transformedsurface layer was found to be larger than in ground samples,indicating that not only stresses, but also the locally devel-oped temperatures during sandblasting were lower. The

T. Kosmacˇ et al. / Dental Materials 15 (1999) 426–433430

Table 3Biaxial flexural strength and Weibull parameters�N � 10� for fine and coarse grained Y-TZP ceramic

Surface treatment Y-TZP creamic Mean flexural strength in MPa (SD) Weibull parameterm (SD) Weibull parameters0 (SD)

As sintered Fine 1021 (89.5) 10.7 (3.2) 1130 (68)Coarse 914 (58.3) 14.9 (4.5) 984 (30)

Dry ground Fine 543 (84.9) 6.2 (1.9) 644 (32)Coarse 724 (153.3) 4.6 (1.4) 907 (90)

Wet ground Fine 642 (116.1) 5.2 (1.6) 785 (55)Coarse 751 (132.3) 4.5 (1.4) 944 (94)

Sandblasted Fine 1239 (146.6) 7.5 (1.7) 1495 (120)Coarse 1224 (119.5) 9.9 (3.0) 1365 (75)

Dryground1 sandblasted Fine 1093 (225.7) 4.5 (1.4) 1378 (138)Coarse 1174 (234.8) 4.5 (1.4) 1480 (148)

Sandblasted1 dry ground Fine 587 (121.8) 4.6 (1.4) 735 (73)Coarse 608 (152.5) 3.9 (1.3) 791 (79)

calculated transformed zone depth, roughly correspondingto the average grain size of sintered Y-TZP ceramic, impliesthat only those tetragonal zirconia grains have transformedinto the monoclinic symmetry during sandblasting, whichare forming the very surface layer. Since these grains are nothydrostatically constrained as those in the bulk of the mate-rial, they can readily transform under the stresses caused byimpacting alumina particles. Although the thickness of thesurface compressive layer formed during sandblasting isvery small, in comparison with the thickness of the testspecimens, it is effective in increasing the strength of theY-TZP materials. The length of surface flaws, which areintroduced by sandblasting, does not seem to exceed largelythe thickness of the compressive surface layer, otherwise thestrength of the material would have been reduced instead ofbeing increased. Recently published results by Peterson etal. [27], who studied strength degradation of various dentalceramic materials upon indentation damage, support thisassumption. According to these authors, the contact damagein Y-TZP, which is closely equivalent to impact damagecaused by sandblasting is manifested in surface-localized,incompletely developed ring cracks which do not extendinto the bulk of the material.

In order to illustrate the counteracting effect of surface

grinding and sandblasting on flexural strength of Y-TZP, aneffective value for the mean critical defect sizeccr for eachgroup was calculated. The mean fracture toughness ofeach material and the corresponding strength values of eachgroup were inserted into Eq. (1), and a penny-shape crackgeometry with its orientation normal to the surface wasassumed by taking 2=p1=2 for the geometric constantw .The calculatedccr values are listed in Table 4. It should beemphasized, that Eq. (1) does not take into account any ofthe residual stress that may exist in the material before beingloaded to fracture. Therefore, except for the two controlgroups, calculated mean critical defect size of ground andsandblasted specimens should be regarded only as an effec-tive length of strength controlling defects, which wouldresult in an equivalent mean strength of the material withoutany residual surface stresses. Bearing this in mind, theresults in Table 4 can be discussed and provide the basisfrom which some general conclusions can be drawn. Forinstance, the calculatedccr value for the as-sintered finegrained Y-TZP is smaller than that in the coarse grainedmaterial, which is a commonly reported observation, andrelates to the inherent flaw population generated duringsintering of fine and coarse powders [28,29]. Surface grind-ing, dry or wet, increases the effective critical defect size bygenerating surface cracks which can be readily observed bySEM examination of fracture surfaces. An example, show-ing the fracture surface of a dry ground fine grained speci-men, is represented in Fig. 3. The site of failure initiationindicated by an arrow in the low magnification (Fig. 3a) wasidentified as a grinding-induced crack extending about50mm from the ground surface into the bulk of the material.With reference to Table 4, grinding-induced cracks extenddeeper into the fine grained Y-TZP than in the coarsegrained Y-TZP. Similarly, greater flaw tolerance of toughercoarse grained TZP ceramics upon indentation wasobserved by Readey et al. [30]. By using water spray duringgrinding, the calculatedccr of dry ground specimens wasreduced by about 30%, indicating lower stresses during

T. Kosmacˇ et al. / Dental Materials 15 (1999) 426–433 431

Table 4Effective mean critical defect size (ccr) obtained from fine and coarsegrained zirconia Y-TZP ceramic

Surface treatment Effective mean critical defect sizeccr

in mm (SD)

Fine Y-TZP Coarse Y-TZP

As sintered 15.2 (3.0) 20.9 (2.7)Dry ground 55.9 (15.7) 37.2 (15.6)Wet ground 41.7 (17.7) 34.4 (18.9)Sandblasted 10.9 (4.5) 11.8 (2.3)Dry ground1 sandblasted 14.8 (6.8) 14.4 (7.6)Sandblasted1 dry ground 50.8 (23) 42.9 (21.7)

Fig. 3. SEM micrograph of the fracture surface of fine grained dry ground Y-TZP: (a) overview, the site of failure is indicated by an arrow; (b) highermagnification of the indicated area. Note that failure originated from a grinding-induced surface crack.

wet grinding. In comparison with Si3N ceramics, a furtherreduction inccr can be expected by using a finer diamondburr [15]. Sandblasting followed by dry grinding appears tobe at least as detrimental surface treatment as dry grindingalone. It results in the highest calculatedccr in the coarsegrained material and next highestccr value in the finegrained Y-TZP, with highest statistical variability in theeffective critical flaw size in both materials.

If, however, dry grinding was followed by sandblasting,the effective mean critical defect size was reduced to one-third of its initial value that is reflected in significantstrengthening of ground Y-TZP ceramics. Microscopicexamination of these samples revealed that a substantialerosive wear through chipping has occurred during sand-blasting which largely leveled the surface of ground mate-rial. The SEM image of an eroded surface shown in Fig. 4also indicates that some of the larger grinding-induced

cracks were removed from the surface during sandblasting.However, since lateral crack chipping is a most prevalentmechanism involved in erosive wear [31], lateral crackswere expected instead, which was later confirmed by fracto-graphic examination. In most cases, failure of these sampleswas initiated from a surface crack linked to subsurfacecracks, as illustrated in Fig. 5. Notice that the critical flawsize, which initiated failure of this fine grained specimen, isevidently larger (50mm) than the effective value forccr

reported in Table 4 for this particular group (14.8mm).The difference between the true and calculated effectiveflaw size is believed to be due to surface compressive stres-ses introduced during sandblasting such that the reducedccr

mainly reflects the contribution of surface compressivelayer to the mean strength of Y-TZP ceramic. This effectwas even more pronounced when the surface of the speci-mens had not been previously damaged by grinding.However, large scatter in strength values, as reflected indecreased Weibull modulus of the sandblasted samples,indicates that during this operation the ceramic surfacewas not only constrained but also damaged. Figs. 4 and 5strongly support this assumption. Under clinical conditions,where the material is exposed to thermal and mechanicalcycling in a chemically active aqueous environment overlong periods, these impact flaws may grow to become stressintensifiers, facilitating fracture initiation at lower levels ofapplied stress. Since coarse grained ceramics exhibit stron-ger crack resistance toughening, it is likely to expect theeffect to be more pronounced in fine grained Y-TZP.

The influence of various mechanical treatments on thereliability of sintered Y-TZP ceramic can best be discussedin terms of the Weibull parameters reported in Table 3. Ifthe clinical failures of Y-TZP dental ceramic were primarilydue to local overloadings on bending, the largest number ofaccidental failures could be expected with the dry ground—fine or coarse grained—material which was sandblastedprior to grinding. By omitting sandblasting prior to grindingand by using water spray during this operation, the perfor-mance of the material should be improved but not to theextent observed on sandblasting of already ground Y-TZPceramic. The formation of a compressive surface layerduring subsequent sandblasting largely compensates forthe grinding-induced surface cracks by pushing the normal-ized strength (s0) of the material toward the higher values.Although, sandblasting does not increase the Weibull modu-lus of the already ground Y-TZP ceramic, the increaseds0

results in acceptable low failure probability of the materialup to the bending stress level of about 500 MPa. Accordingto the results presented in Table 3, the as-sintered coarse andfine grained material can be exposed to approximately700 MPa bending stress level without any serious risk offailure. Since, coarse grained Y-TZP is also less susceptibleto grinding induced defects, it appears that tougher coarsegrained material might be suggested for dental applicationof zirconia ceramic requiring surface treatment, in spite ofsomewhat lower mean strength before surface treatments.

T. Kosmacˇ et al. / Dental Materials 15 (1999) 426–433432

Fig. 4. SEM micrograph of the surface of fine grained dryground1 sandblasted Y-TZP showing remaining grinding scratches andsharp sandblasting scures. The predominant mechanism of materialremoval during sandblasting was surface chipping.

Fig. 5. SEM micrograph of the fracture surface of fine grained dryground1 sandblasted Y-TZP. Failure originated from a 50mm deepsurface pit.

It should be pointed out, that the strength of damagedzirconia is still higher than the strength of other full-ceramicsystems even without any surface treatment [32,33]. Underclinical conditions, however, dental restorations are exposedto stress and chemical interactions that are more complexand deleterious than those used in our strength degradationtests. It is, therefore, reasonable to expect first accidentalfailures to occur at stress levels lower than those reported inour study.

References

[1] Meyenberg KH, Luthy H, Scharer P. Zirconia posts: a new all-cera-mic concept for nonvital abutment teeth. J Esthet Dent 1995;7:73–80.

[2] Paul S, Scha¨rer P. Adha¨sivaufbauten fu¨r Vollkeramik-Kronen.Schweiz Monatsschr Zahnmed 1996;106:368–374.

[3] Luthardt R, Herold V, Sandkuhl O, Reitz B, Knaak JP, Lenz E.Kronen aus Hochleistungskeramik. Dtsch Zahnarztl Z1998;53:280–285.

[4] Wohlwend A, Studer S, Scha¨rer P. The zirconium oxide abutment: anall-ceramic abutment for the esthetic improvement of implant super-structures. Quintessence Dent Technol 1997;1:63–74.

[5] Keith O, Kusy RP, Whitley JQ. Zirconia brackets: an evaluation ofmorphology and coefficients of friction. Am J Orthod DentofacialOrthop 1994;106:605–614.

[6] Subbarao EC. Zirconia—an overview. Adv Ceram 1981;3:1–24.[7] Wolten GM. Diffusionless phase transformations in zirconia and

hafnia. J Am Ceram Soc 1963;46:418–422.[8] Garvie RC, Hannink RH, Pascoe RT. Ceramic steel? Nature

1975;258:703–704.[9] Lawn BR. Fracture of brittle solids, 2nd. Cambridge: Cambridge

University Press, 1993.[10] Lange FF. Powder processing science and technology of increased

reliability. J Am Ceram Soc 1989;72:99–104.[11] Pascoe RT, Garvie RC. In: Fulrath RM, Pask JA, editors. Ceramic

microstructures, Boulder, CO: Westview Press, 1977. pp. 774–785.[12] Gupta TK. Strengthening by surface damage in metastable tetragonal

zirconia. J Am Ceram Soc 1980;63:117–121.[13] Claussen N. Transformation toughened ceramics. Z Werkstofftechn

1982;13:138–147.[14] Green DJ. A technique for introducing surface compression into

zirconia ceramics. J Am Ceram Soc 1983;66:C178–C179.[15] Erauw JP, Hendrix W, Van Hoof E, Breder K, Ferber MK. Machining

of silicon nitride: process parameters—mechanical properties rela-tionship. In: Niihara K, Kanzaki S, Komeya K, Hirano S, MorinagaK, editors. Proceedings of the Sixth International Symposium on

Ceramic Materials and Components for Engines: October 19–23,1997, Arita, Japan, 1997.

[16] Kern M, Wegner SM. Bonding to zirconia ceramic: adhesion methodsand their durability. Dent Mater 1998;14:64–71.

[17] Swain MV. Grain-size dependence of toughness and transformabilityof 2 mol% Y-TZP ceramics. J Mater Sci Lett 1986;5:1159–1162.

[18] Garvie RC, Nicholson PS. Phase analysis in zirconia systems. J AmCeram Soc 1972;55:303–305.

[19] Kosmac T, Wagner R, Claussen N. Determination of transformationdepths in ceramics containing tetragonal ZrO2. J Am Ceram Soc1981;64:C72–C73.

[20] Chantikul P, Anstis GR, Lawn BR, Marshall DB. A critical evaluationof indentation techniques for measuring fracture toughness. II.Strength method. J Am Ceram Soc 1981;64:539–543.

[21] Wachtman JB, Capps W, Mandel J. Biaxial flexure tests of ceramicsubstrates. J Mater 1972;7:188–194.

[22] Ritter JE, Bandyopadhyay N, Jakus K. Statistical reproducibility ofdynamic and static fatigue experiments. Ceram Bull 1981;60:798–806.

[23] Gross V, Swain MV. Mechanical properties and microstructure ofsintered and isostatically pressed yttria partially stabilized zirconia(Y-PSZ). J Aust Ceram Soc 1986;22:1–12.

[24] Urabe K, Nakajima A, Ikawa H, Udagava S. Characteristic micro-structure of CeO2–Y2O3 dopped tetragonal zirconia polycrystals. AdvCeram 1988;24:345–355.

[25] Virkar AV, Matsumoto RLK. Ferroelastic domain switching as atoughening mechanism in tetragonal zirconia. J Am Ceram Soc1986;69:C224–C226.

[26] Swain MV, Hannink RHJ. Metastability of the martensitic transfor-mation in a 12 mol% ceria–zirconia alloy: grinding studies. J AmCeram Soc 1989;72:1358–1364.

[27] Peterson IM, Pajares A, Lawn BR, Thompson VP, Rekow ED.Mechanical characterization of dental ceramics by herzian contacts.J Dent Res 1998;77:589–602.

[28] Rice RW. Strength grain size effects in ceramics. Proc Brit Ceram Soc1972;20:205–207.

[29] Seidel J, Claussen N, Ro¨del. Reliability of alumina ceramics: effect ofgrain size. J Eur Ceram Soc 1995;15:395–404.

[30] Readey MJ, Mc Callen CL, Mc Namara PD, Lawn BR. Correlationsbetween flaw tolerance and reliability of ceramics. J Mater Sci1993;28:2748–2752.

[31] Ritter JE. Particle impact damage of engineering ceramics. In: RitterJE, editor. Erosion of ceramic materials, Trans Tech, 1992. pp. 555–578.

[32] Hondrum SO. A review of strength properties of dental ceramics. JProsthet Dent 1992;67:859–865.

[33] Kappert HF. Dental materials: new ceramic systems. Trans Acad DentMater 1996;9:180–199.

T. Kosmacˇ et al. / Dental Materials 15 (1999) 426–433 433