Sz

VNa

b

c

Kd

a

A

R

R

2

A

K

C

F

W

C

F

Z

f

h0

d e n t a l m a t e r i a l s 3 1 ( 2 0 1 5 ) 77–87

Available online at www.sciencedirect.com

ScienceDirect

jo ur nal home p ag e: www.int l .e lsev ierhea l th .com/ journa ls /dema

low cooling protocol improves fatigue life ofirconia crowns

itor G. Paulaa, Fabio C. Lorenzonid, Estevam A. Bonfantea,∗,elson R.F.A. Silvab, Van P. Thompsonc, Gerson Bonfantea

Department of Prosthodontics, University of São Paulo – Bauru College of Dentistry, Bauru, SP, BrazilDepartment of Restorative Dentistry, Federal University of Minas Gerais, UFMG, Belo Horizonte, MG, BrazilDepartment of Biomaterials, Biomimetics and Biophotonics, King’s College London, Guy’s Hospital, London, UnitedingdomDepartment of Prosthodontics, UNIGRAN University, Dourados, MS, Brazil

r t i c l e i n f o

rticle history:

eceived 2 September 2013

eceived in revised form

2 December 2013

ccepted 27 October 2014

eywords:

eramics

atigue

eibayes

ooling rate

ractography

irconia

a b s t r a c t

Objective. To compare the fatigue life and damage modes of zirconia crowns fabricated with

and without framework design modification when porcelain veneered using a fast or slow

cooling protocol.

Methods. Composite resin replicas of a first molar full crown preparation were fabricated.

Zirconia copings were milled as conventional (0.5 mm even thickness, Zr-C, n = 20,) or mod-

ified (lingual margin of 1.0 mm thickness, 2.0 mm height connected to two proximal struts

of 3.5 mm height, Zr-M, n = 20). These groups were subdivided (n = 10 each) according to the

veneer cooling protocol employed: fast cooling (Zr-CFast and Zr-MFast) and slow cooling

(Zr-CSlow and Zr-MSlow). Crowns were cemented and fatigued for 106 cycles in water. The

number of cycles to failure was recorded and used to determine the interval databased 2-

parameter probability Weibull distribution parameter Beta (ˇ) and characteristic life value

Eta (�).

Results. 2-parameter Weibull calculation presented = 5.53 and = 4.38 for Zr-MFast and Zr-

CFast, respectively. Slow cooled crowns did not fail by completion of 106 cycles, thereby

Weibayes calculation was applied. Increased fatigue life was observed for slow cooled crowns

compared to fast cooled ones. Groups Zr-MFast and Zr-MSlow presented no statistical differ-

ence. Porcelain cohesive fractures were mainly observed in fast cooled groups. Slow cooled

crowns presented in some instances inner cone cracks not reaching the zirconia/veneer

interface.

Significance. Improved fatigue life in tandem with the absence of porcelain fractures were

observed in slow cooled crowns, regardless of framework design. Crowns fast cooled chiefly

failed by porcelain cohesive fractures.

© 2014 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved.

∗ Corresponding author at: Alameda Octávio Pinheiro Brisola 9-75, 17.01ax: +55 14 3234 5805.

E-mail address: estevamab@gmail.com (E.A. Bonfante).

ttp://dx.doi.org/10.1016/j.dental.2014.10.005109-5641/© 2014 Academy of Dental Materials. Published by Elsevier L

2-901 Bauru, SP, Brazil. Tel.: +55 14 98153 0860;

td. All rights reserved.

l s 3

78 d e n t a l m a t e r i a

1. Introduction

Over the past ten years several in vitro [1–5] and clinical stud-ies [6,7] have attempted to document and understand thenature of porcelain veneer cohesive fractures in all-ceramicrestorations, specially yttria-tetragonal zirconia polycrystals(Y-TZP) [6,7]. Besides issues related to manufacturing process(i.e., porosities and micro-structural defects) [8], chipping hasbeen suggested to be strongly related with; residual thermalstress [9–12], coefficient of thermal expansion mismatches [1],poor framework design [3,4,13–15], and others [12].

Changes in the porcelain physical properties (during thetransition from the visco-elastic to solid state) when it iscooled down from the sintering temperature to room tem-perature take place in every firing cycle [16,17]. As a result,non-uniform residual thermal stresses may be producedthroughout the porcelain veneer as a function of coolingrates [18,19]. Such residual stress may be induced within theporcelain veneer either by thermal expansion mismatchesor by the temperature gradients produced during cooling orboth [12,17]. Similarly to tempered glass, during a fast cool-ing process the outer crown surface becomes solid first andstarts to contract inducing a compressive stress. However,while the inner surface is still in a visco-elastic state andat high temperature, leading to high tensile stress when thesystem reaches room temperature [12,17,19]. This process isexpected given that the porcelain veneer inner bulk remainsin a temperature above its glass-transition temperature (Tg),leading to the gradient temperature (difference between theinner and outer temperature) [12]. As a consequence, resid-ual thermal stresses are held within the porcelain veneer[9,12,20], also known as tempering. In addition, Y-TZP shows alow thermal diffusivity (thermal conductivity of 2 W m−1 K−1

whereas metals vary from 100 to 300 W m−1 K−1) [21,22],which may increase the thermal gradient intensity withinveneer porcelain [9,12,20,23]. Therefore, the multifaceted roleof the above factors, has been speculated as potential sourcesrelated to high Y-TZP chipping rates reported in the literature[5,8,10,12,23].

Recently, efforts have been devoted to the reduction ofresidual thermal stresses in Y-TZP restorations by managingthe rates of cooling [9,10,24,25]. It has been reported that over-all thermal gradients within the porcelain veneered to Y-TZPdecreased when a slow cooling protocol was used [23,26]. Also,thermodynamic analysis performed in molar crowns showedthat different results may be achieved as a function of coolingprotocols [23]. It has been shown that slow cooling (the furnacedoor opened when the temperature was below the porcelainveneer Tg) produced a small temperature variation betweenthe inner and outer surfaces of crowns and, more important,it decreased the residual thermal stresses within the veneerlayer [23]. However, to date few attempts regarding the influ-ence of cooling rate on the fatigue life of Y-TZP crowns areavailable.

Most manufacturers of porcelain veneers for Y-TZP restora-

tions have modified their firing guidelines, advocating thattechnicians adopt a slow cooling rate protocol. In 2009 forexample, one manufacturer recommended that the furnacewas opened soon after the end of the firing cycle [27], and

1 ( 2 0 1 5 ) 77–87

a year later it was modified and crowns should be kept intothe closed furnace up to temperature reaching 450 ◦C, slightlybelow the Tg (490 ◦C) of the veneer porcelain [28].

Another topic that has frequently been revisited is frame-work design modifications in Y-TZP restorations [4,14,29].Clinical evaluations have suggested that porcelain chippingis frequently associated with areas without appropriate porce-lain support (proximal areas) [14,15,30]. This empirical conceptrelies on the fact that the porcelain veneer should have aneven thickness, of no more than 2 mm, and be supported bythe core [31]. Although framework design modifications haveshown to decrease chipping rates in clinical trials [15,32,33]no control groups (i.e., even thickness framework) have beenincluded, hindering a more comprehensive understandingbetween framework design modification and chipping [15,32].

In light of the aforementioned information, this studysought to evaluate the fatigue life and failure modes of Y-TZPmolar crowns fabricated with two different porcelain veneercooling protocols with and without framework design modifi-cation. The following null hypotheses were tested: (1) coolingprotocols during glazing will not improve the fatigue life of Y-TZP systems and (2) framework design modification will notimprove the fatigue life of Y-TZP crown systems, irrespectiveof cooling rate.

2. Materials and methods

2.1. Sample preparation

A plastic maxillary first molar positioned in a mannequin(Plastic mannequin tooth – MOM, Marilia, SP, Brazil) was pre-pared to receive a full crown by reducing the axial walls by1.5 mm, the occlusal surface by 2.0 mm and shoulder margin(1.2 mm of thickness) with rounded internal angles. After-wards, 40 composite resin replicas were obtained from a vinylpolysiloxane impression (Express – 3M ESPE, St. Paul, MN,USA) of the prepared tooth, which was incrementally (2.0 mmof thickness) packed with composite resin (Z100 – 3M ESPE,St. Paul, MN, USA) and light-cured (Ultralux, Dabi Atlante,Ribeirão Preto, SP, Brazil) according to manufacturer’s instruc-tions. All resin tooth replicas were stored in distilled waterfor at least 30 days to minimize dimensional changes duringtesting [34]. Each replica was vertically positioned in 25 mmdiameter PVC tubes and potted with acrylic resin (Jet, ClássicoArtigos Odontológicos, São Paulo, SP, Brazil). The preparationfinish line was maintained 2 mm above the potting sur-face. Stone dies were acquired from polyether impressions(Impregum F – 3M-ESPE, St. Paul, MN, USA) of each resin-toothreplica and sent to a dental laboratory.

Stone dies were randomly divided in 2 groups (n = 20, each)according to framework design, as: conventional (Zr-C) ormodified design (Zr-M). Subsequently, each group was subdi-vided in two subgroups (n = 10) according to thermal coolingprotocol as: fast cooling protocol (Zr-CFast and Zr-MFast) andslow cooling protocol (Zr-CSlow and Zr-MSlow).

Frameworks were milled from pre-sintered Y-TZP blocks(e.max ZirCAD, shade MO 0, block size 15C – Ivoclar Vivadent,Schaan, Liechtenstein) and sintered in the furnace (Sintra-mat, Ivoclar Vivadent AG, Schaan, Liechtenstein) at 1500 ◦C

d e n t a l m a t e r i a l s 3 1 ( 2 0 1 5 ) 77–87 79

Fig. 1 – Proximal view of (a) standard and (b) modified core designs before veneering with porcelain.

Table 1 – Programat® EP 3000 – Firing program table(Ivoclar Vivadent) – March 2010 [31].

Program ZirLiner e.Max Ceram

Stand-by temperature 403 ◦C 403 ◦CStand-by time 4 min 4 minTemperature increase 40 ◦C/min 40 ◦C/minHolding temperature 960 ◦C 750 ◦CHolding time 1 min 1 minVacuum on 450 ◦C 450 ◦CVacuum off 959 ◦C 749 ◦C

◦ ◦

fn1oa

CLmhm(lwis–ptstIwt

•

Opening furnace 960 C 750 CCooling rate – –

or 8 h. Zr-C cores were manufactured with an even thick-ess of 0.5 mm. For Zr-M cores, a lingual margin (thickness of.0 mm with 2.0 mm height) connected to two proximal strutsf 3.5 mm height were added to an even thickness of 0.5 mm,s described previously [3] (Fig. 1).

All Y-TZP cores were veneered with porcelain (IPS e.maxeram, shade Transpa Clear – Ivoclar Vivadent AG, Schaan,iechtenstein) and sintered in a ceramic furnace (Progra-at EP3000 – Ivoclar Vivadent, Schaan, Liechtenstein). The

and-layering process was performed in accordance with theanufacturer’s recommendation, starting with one liner firing

ZirLiner – Ivoclar Vivadent AG, Schaan, Liechtenstein) fol-owed by firing two layers of porcelain. Finally, the crownsere manually finished and polished before the glaze fir-

ng. The firing schedule of liner and veneering porcelain arehown in Table 1 and Fig. 2A. A silicone matrix (Zetalabor

Zhermack, Badia Polesine, Rovigo, Italy) of the unpreparedlastic mannequin tooth was used to guide and standardizehe porcelain veneer contour. The occlusal and axial surfaceshowed approximately 1.5 and 1.0 mm of thickness, respec-ively. The glaze layer (Glaze Paste and Stain Liquid Long Life,PS e.Max Ceram – Ivoclar Vivadent AG, Schaan, Liechtenstein)as applied onto the porcelain surface and treated according

o one of the following cooling protocols (Fig. 2B and Table 2):

Fast cooling: after the end of the glaze firing cycle, the fur-nace head was opened. Then, crowns were immediatelytaken out from the mesh-tray to the bench, leading to cool

Fig. 2 – (A) Programat® EP 3000 – Firing program (IvoclarVivadent) – March 2010 [28]. (B) Glaze firing program.

80 d e n t a l m a t e r i a l s 3

Table 2 – Glaze firing program table.

Program Fast cooling Slow cooling

Stand-by temperature 403 ◦C 403 ◦CStand-by time 6 min 6 minTemperature increase 60 ◦C/min 60 ◦C/minHolding temperature 725 ◦C 725 ◦CHolding time 1 min 1 minVacuum on 450 ◦C 450 ◦CVacuum off 724 ◦C 724 ◦C

◦ ◦

cone cracks reaching the interface reflect the porcelain’s fail-

Opening furnace 725 C 50 CCooling rate – 10 ◦C/min

at room temperature. This thermal cycle followed manufac-turer’s recommendation published in 2009 [27].

• Slow cooling protocol: the furnace had the cooling rate set to10 ◦C/min and crowns were kept into the furnace until thetemperature reached 50 ◦C, after which the furnace headwas opened.

The glazed crowns were cemented to the aged resin-composite tooth replicas using a dual cure resin cement (RelyX ARC – 3M-ESPE, St. Paul, MN, USA) following the manu-facturer’s instructions. No treatment was made in the Y-TZPcementation surface. Prior to mechanical testing, sampleswere incubated in distilled water at 37 ◦C for at least 24 h.

2.2. Fatigue test

Fatigue test was carried out in a fatigue machine (Model MSFM– Elquip – São Carlos, SP, Brazil) in r-ratio mode at a 30–300 Nload range for 106 cycles or until failure under distilled water(37 ◦C) [3]. As the indenter was in contact with the specimen,the initial contact load of 30 N was applied without abrasion,attrition or impact, steadily increasing to 300 N and decreasingback to 30 N. The load was applied through a spherical steelindenter (3.18 mm radius) on the center of the occlusal sur-face, at 2 Hz. Our attempt was to simulate the most typicalmolar relationship in natural dentition (Angle class I) wherethe mesiolingual cusp of the maxillary first molar is situatedin the central fossa area of the mandibular first molar [35].After a predetermined number of cycles (125,000) the test wasinterrupted for crown surface damage accumulation inspec-tion under stereomicroscopy (Leica Zeiss MZE, Mannheim,Germany). Specimens were then repositioned in the machinefor an additional 125,000 cycles and inspections until comple-tion of 106 cycles or failure.

2.3. Data analysis

The failure during fatigue was recorded as a function of thenumber of cycles. This information was used to determinethe interval databased 2-parameter probability Weibull dis-tribution for Zr-CFast and Zr-MFast (Weibull 7++; Reliasoft,Tucson, AZ, USA). Weibull shape parameter Beta (ˇ) and char-acteristic value Eta (�) (here the number of stress cycles (n) atwhich 62.3% of the specimens would fail) were determined

employing 90% confidence levels for upper and lower lim-its where applicable. If either one of groups displayed nofailures, a Weibayes [36] calculation was employed based on

1 ( 2 0 1 5 ) 77–87

the of the counter group’s distribution. Weibayes is a one-parameter Weibull analysis technique developed to “solveproblems when traditional Weibull analysis has large uncer-tainties or cannot be used because there are no failures” [36].“A basic assumption governing the accuracy of Weibayes anal-ysis and its associated test regimens is that the value of theWeibull shape parameter (ˇ) is known or can be reasonablyestimated”. “Knowledge of can be derived from historicalfailure data, prior experience, or from engineering knowledgeof the physics of the failure”. Given ˇ, an equation may bederived using the method of maximum likelihood to deter-mine the characteristic life, �, as follows [36,37]:

� =[

N∑i=1

Tiˇ

r

] 1ˇ

(1)

where Ti, test time for each sample; r, number of failures; N,sample size; �, characteristic life of the cumulative densityfunction (CDF) = 63.2%; ˇ, Weibull shape parameter.

A Weibull distribution is derived from � (determined by Eq.(1)) using the assumed ˇ. The 2-parameter Weibull probabilityresult was plotted using the counter group Weibayes value forpairwise comparisons. If two or more groups showed no fail-ures, the Weibayes lines were plotted and a probability Weibullgraphic generated.

2.4. Failure mode characterization

The criteria for failure were inner cone cracks reachingthe Y-TZP framework, porcelain veneer cohesive fractureor Y-TZP core fracture. Representative fractured surfaceswere characterized under polarized-light microscopy (MZ-APO stereomicroscope, Carl Zeiss MicroImaging, Thornwood,NY, USA) followed by Scanning Electron Microscopy (SEM)(Model S-3500N; Hitachi, Japan). If no fractures were notedafter completion of final fatigue interval, the specimen wasconsidered as suspended. Subsequently, to characterize thesubsurface damage, suspended crowns were embedded inepoxy resin (Resina Epoxi RD6921, Redelease, São Paulo,Brazil), sectioned and then serially polished along the plane ofcrack extension with silicon carbide papers (320, 400, 500, 600,1200, 2000 and 2500 grits), under copious water irrigation. Thedamage accumulation was inspected and registered understereomicroscopy (Leica Zeiss MZE, Mannheim, Germany).

3. Results

During fatigue testing, all crowns fast cooled, from Zr-CFastand Zr-MFast groups, failed before completion of 1,000,000cycles chiefly by porcelain cohesive fracture. Three crownswith conventional framework and five from the modified fastcooled groups did not present chippings. However, the sub-surface damage inspection of these crowns revealed innercone cracks reaching the core/veneer interface. While inner

ure to resist crack growth, they do not represent a clinicalfracture per se, only the porcelain fractured crowns were con-sidered for further analysis (n = 7 from the Zr-CFast and n = 5

d e n t a l m a t e r i a l s 3 1 ( 2 0 1 5 ) 77–87 81

Fig. 3 – (a) Probability Weibull 2P plot showing probability line for Zr-CFast and Weibayes line projection of group Zr-CSlow.Note that 90%CB calculation for Zr-CFast did not allow confidence intervals. (b) Probability Weibull 2P plot showingprobability line calculation of Zr-MFast and Weibayes probability line projection of group Zr-MSlow. (c) Probability Weibullplot showing the Zr-CSlow and Zr-MSlow Weibayes projection lines. Note the different inclination of the probability lines asa d =

fcac

result of the different Beta values: = 4.38 for Zr-MSlow an

rom the Zr-MFast). Conversely, crowns subjected to the slowooling protocol, regardless of framework design, (Zr-CSlownd Zr-MSlow) were suspended since they did not fail at theompletion of 106 cycles. Results of the Weibull calculation

5.53 for Zr-CSlow.

are shown in Fig. 3. For the results distribution, the groupswere plotted based on the core design varying the cooling rateprocess as described in materials and methods. 2-ParameterWeibull calculation was performed for Zr-CFast and Zr-MFast.

82 d e n t a l m a t e r i a l s 3 1 ( 2 0 1 5 ) 77–87

Fig. 4 – (a) Light-polarized microscope and SEM images of a representative failed Zr-MFast crown. Palatal view of cohesiveporcelain fracture shows that the crack involved the lingual cusp. (b) SEM lingual view displaying the indentation area(asterisk) and the dotted arrows shows the crack propagation toward the cervical and proximal areas. (c) and (d) Theconcave portion of the arrest lines (pointer) points to the origin (asterisk). Continued propagation toward the margins issuggested by the series of hackles (bowshot). (e) and (f) Magnification of wake-hackles (pointers) corroborating the origin

(whi

and direction of crack propagation along with twist-hackles

Groups Zr-CSlow and Zr-MSlow did not present any failure,and therefore Weibayes calculation was performed [36]. In ourstudy we used the value from the respective counter group’sdesign and ran Weibayes calculation. 2-parameter Weibullcalculation presented = 5.53 and = 4.38 for Zr-MFast and Zr-CFast respectively. The Weibayes calculation using the same

values for the respective counter presented a shift to theright of the Weibull probability lines (longer fatigue life). The 2-parameter Weibull calculation for group Zr-CFast did not allow90% confidence intervals calculation as a result of the distri-

bution of the failed specimens. However there is a significantshift to the right of the Weibayes probability line for its counterpart group. Groups Zr-MFast and Zr-MSlow presented no sta-tistical difference as a result of overlap of the upper limit of

te arrows).

the confidence bounds. Fig. 3 also presents the Weibayes prob-ability line calculations for groups Zr-MSlow and Zr-CSlow.

3.1. Failure modes

The chief failure mode observed for crowns subjected to thefast cooling protocol was porcelain veneer cohesive fracture,regardless of framework design (Zr-CFast, Zr-MFast), with-out exposing the Y-TZP surface. Of special interest was thefact that smaller chippings were consistently observed for

Zr-MFast relative to Zr-CFast group. In the Zr-MFast group,chippings were generally confined to the lingual cusp (Fig. 4),whereas chip-off fractures extending to the proximal sur-faces were observed in Zr-CFast crows (Fig. 5). Qualitative

d e n t a l m a t e r i a l s 3 1 ( 2 0 1 5 ) 77–87 83

Fig. 5 – (a) Light-polarized microscope image of a Zr-CFast crown shows the extension of the fractured porcelain veneer.Note that a layer of porcelain remained onto the Y-TZPs framework surface. (b) SEM lingual overview of the fractured crownand indentation area (asterisk) where the fracture initiated and propagated toward margins and proximal areas (dottedarrows). (c) Magnification of (b) showing the presence of a series of wake-hackles (pointers) confirming the fracture originand direction of the crack propagation (dotted arrow). (d) A series of twist-hackles (white arrows) that strongly suggests thep direc

fSilcghc

fectdgf

4

Talcons

resence of residual thermal stress and further confirms the

ractography performed in polarized-light microscopy andEM showed the presence of quasiplastic deformation at the

ndentation area. Telltale fractographic marks including hack-es, wake hackles, and arrest lines suggesting the direction ofrack propagation from the indentation area toward the mar-ins of the fractured porcelain veneer were observed. Twistackles were only identified in those groups subjected to fastooling.

Crowns subjected to the slow cooling protocol of bothramework designs (Zr-CSlow and Zr-MSlow) were consid-red suspended since they did not present failures at theompletion of the 106 cycles. The subsurface damage inspec-ion of these specimens revealed inner cone cracks as theominant crack system mechanism in all samples from bothroups, although they did not reach the core/porcelain inter-ace (Fig. 6). No radial crack and bulk fractures were observed.

. Discussion

he present study evaluated the effect of cooling processesnd framework designs on Y-TZP molar crowns fatigueife. The absence of statistical difference in fatigue life in

rowns submitted to the slow cooling protocol, regardlessf framework design, lead us to reject the first postulatedull hypothesis. The second null hypothesis was acceptedince framework design modification did not improve fatigue

tion of crack propagation toward the margin of the fracture.

life within either the fast or slow cooled crowns. Howeverwhen intergroup comparisons were made, the relevance of amodified framework should not be overlooked since the Zr-MFast crowns were not statistically different from slow cooledcrowns of any framework design. Nevertheless, it appears thatcooling protocols played a major role in fatigue life in thecurrent study. As expected from previous studies [6,9,38], noY-TZP core fractures were detected. When porcelain veneerfractures occurred, they were always cohesive within theporcelain veneer, without exposing the Y-TZP core [3,5,39–42].

Crowns cooled slowly did not present porcelain chippingswithin the established limit of one million fatigue cycles andthe only observed failure mode was near-field damage (innercone crack), in both framework designs. It is likely that suchfatigue performance improvement may be explained by areduction in residual stress entrapment within the veneerlayer [16,20,23]. A recent investigation evaluated the influ-ence of cooling protocols on thermal stress formation in Y-TZPmolar crowns showing that when samples were cooled slowlythe outer porcelain veneer surface did not become solid first(due to the absence of thermal gradient) allowing residualthermal stress relaxation by viscosity flow [23]. As fatiguetesting was suspended at one million cycles to allow for dam-

age inspection of groups within the same range of cycles,it must be acknowledged that extended fatigue could haveprovided a more clinically realistic failure scenario, althoughincreasing testing time to an unknown extent. In addition, as

84 d e n t a l m a t e r i a l s 3 1 ( 2 0 1 5 ) 77–87

Fig. 6 – (a) Occlusal light-polarized micrograph shows the indenter contact area on the porcelain surface and the resultingquasiplastic deformation on a Zr-MSlow crown. (b) Magnification of the square area showed in (a), reveals the presence ofquasiplatic deformation, which is a coalescing microcrak involving the indentation area (bowshot) after 106 cycles. (c)Bucco-lingual section of the same crown, polished and examined throughout the damaged area, showing inner cone crackformation immediately below the indentation area, not reaching the zirconia/veneer interface (d) Magnification of the innercone crack confirming its extension. (e) Bucco-lingual section micrograph of a polished Zr-CSlow crown showing theindentation area in the blue square and in (f) its magnification revealing inner cone cracks (pointer) formed immediatelybellow the indentation area, also not reaching the interface. R = resin composite; Y-TZP = zirconia framework; V = veneering

porcelain.

sliding-fatigue was omitted, porcelain fractures revealed ori-gin from the occlusal surface, but not related to surfaceroughness due to wear, as reported in clinical fractogra-phy [43] and laboratory studies involving such methodology[2,44,45,47].

Regarding the slow cooling protocol used in this study,it must be emphasized that it is experimental for havingextended the slow cooling process below the Tg, in contrast to

what is currently recommended by most manufacturers (slowcooling in the last stain/glaze firing cycle until Tg). Whereasthe maintenance of the crowns into the furnace until 50 ◦Cdemanded additional laboratory time, it may be important to

reduce thermal gradients [12,23,46]. A comprehensive eval-uation of stress distribution within the veneering ceramicas a function of depth toward the substructure material hasshown that zirconia veneered samples present tensile stressesaround 0.5 mm from the surface with magnitude varyinginversely to the cooling rate [26].

Y-TZP crowns cooled fast resulted in similar fatigue life,irrespective of framework design, but remarkable differences

were observed in porcelain cohesive fracture sizes. Whereasfractures in the modified Y-TZP framework were confined tolingual cusp, those in the conventional designed were moreextensive often involving both approximals in addition to the

s 3 1

bpbSueacmarram

pptsdp

avoasthobtbai

ftgip(lSttoao

pddsacaioa

r

d e n t a l m a t e r i a l

uccal cusp. Despite such differences, from a clinical stand-oint only crowns with modified framework would potentiallye repaired, although ideally both would require replacement.uch an observation questions the suitability of this partic-lar framework design modification since it results in anven porcelain thickness only on the lingual side. Perhaps, annatomic core design that provides an even thickness in allusps and proximal areas could be more relevant as recom-ended by several manufacturers. A recent study shows that

n anatomical support of the veneering ceramic resulted ineduced veneer chip size fractures and significantly increasedeliability with press and hand-layer veneering ceramics andppears to be a promising tool to optimize long-term perfor-ance of veneered Y-TZP crowns in clinical application [29].One important fractographic finding was the consistent

resence of twist hackles in the groups cooled by the fastrotocol. Although it was not our aim to perform residualhermal stress measurements, the presence of twist hackles,trongly suggest the presence of significant thermal stress gra-ients [47], which were induced in our study by the fast coolingrotocol [16,23,46].

Our Weibayes use of an assumed beta value means thatlways there is a “risk” of using a beta higher than the “true”alue [37]. In this context, the impact of a higher beta is beingverly optimistic in the interpretation of the demonstrate reli-bility and the possible exclusion of early failures. In ourtudy the beta value employed is based upon the value fromhe respective counter group (fast cooled), which displayed aigher number of failures than the slow cooled groups. More-ver, a way to investigate the effect of changes in an assumedeta is by assuming two beta values (two beta lines), wherehe higher the cumulative density function (CDF) level of theeta line intersection point, the greater the risk of reachingn overly optimistic conclusion for a fixed By life below thatntersection point [37].

Also, if two beta values were assumed, one as calculatedrom results of the counter group and the other reduced bywo units, one graphic including both beta lines would beenerated. For a given N = 10 (based on our sample size), thentersection of two beta lines is known as beta intersectionoint and it most likely occurs at approximately CDF of 10%

here the test time on each of the “N” samples is equal). For aarger “N”, the beta intersection point occurs at less than 10%.uch assumption is based on a general rule for beta intersec-ion point where the smaller the Weibayes test sample size,he higher the CDF value at which the beta intersection pointccurs [37]. Thus, our limited number of samples would allow

lower beta intersection point close to the B10 life where 10%f samples would fail.

It is acknowledged that several damage modes can com-ete during the fatigue test such as radial cracks (far-fieldamage) and inner, outer, and partial cone cracks (near-fieldamage) [48–50]. The failure characterization by polishinghowed that some of the suspended samples (three ZR-CFastnd four Zr-MFast) presented inner cone cracks reaching theore/veneer interface. Because from a clinical perspective the

bsence of external porcelain fracture, as in the presence ofnner cone cracks reaching the interface, leads to a diagnosisf crown survival, they were not deemed as failures for datanalysis. With continued load cycling it would likely evolve to

( 2 0 1 5 ) 77–87 85

a cohesive fracture from the extension of the inner cone crack-ing [49–52]. Suspended crowns subjected to the slow coolingprotocol showed cone cracks formation virtually confined tothe quasiplastic zone beneath the indenter and therefore notreaching the Y-TZP framework and are much less likely to leadto cohesive failure with continued loading.

As observed in clinical trials, the proximal region has beenregarded as an area specially affected by chipping [6,14,30,53].In a natural occlusal scheme the mesio bucal cusp of the firstmandibular molar may contact the marginal ridges of bothsecond premolar and first maxillary molar [54,55]. This sug-gests that future fatigue testing in the proximal areas may beof clinical relevance.

5. Conclusions

The first postulated null hypothesis which stated that cool-ing protocols during glazing would not improve the fatiguelife of Y-TZP crowns was partially rejected. Decreased fatiguelife was observed only for crowns fast cooled of conventionalframework compared to any slow cooled crowns, whereasmodified framework fast cooled crowns were not signifi-cantly different from slow cooled ones. The second nullhypothesis was accepted since framework design modifica-tion did not improve fatigue life within either the fast orslow cooled crowns. Slow cooling protocol is recommended formore predictable mechanical performance of layered-zirconiarestorations.

Acknowledgements

We thank Dr. Raphael Piazzarollo Loureiro for his input onceramic processing engineering and the dental technician JoséCarlos Romanini for crown fabrication. This study was sup-ported, in part, by CNPq, and FAPESP Grant 2010/06152-9. Theauthors acknowledge Hitachi S3500N SEM imaging that wasmade possible by the New York University College of Den-tistry’s cooperative agreement with the NIH/NIDCR.

e f e r e n c e s

[1] Bonfante EA, Rafferty B, Zavanelli RA, Silva NR, Rekow ED,Thompson VP, et al. Thermal/mechanical simulation andlaboratory fatigue testing of an alternative yttria tetragonalzirconia polycrystal core-veneer all-ceramic layered crowndesign. Eur J Oral Sci 2010;118:202–9.

[2] Coelho PG, Silva NR, Bonfante EA, Guess PC, Rekow ED,Thompson VP. Fatigue testing of two porcelain-zirconiaall-ceramic crown systems. Dent Mater 2009;25:1122–7.

[3] Lorenzoni FC, Martins LM, Silva NR, Coelho PG, Guess PC,Bonfante EA, et al. Fatigue life and failure modes of crownssystems with a modified framework design. J Dent2010;38:626–34.

[4] Silva NR, Bonfante EA, Rafferty BT, Zavanelli RA, Rekow ED,Thompson VP, et al. Modified Y-TZP core design improves

all-ceramic crown reliability. J Dent Res 2011;90:104–8.

[5] Silva NR, Bonfante EA, Zavanelli RA, Thompson VP, FerenczJL, Coelho PG. Reliability of metalloceramic andzirconia-based ceramic crowns. J Dent Res 2010;89:1051–6.

l s 3

86 d e n t a l m a t e r i a

[6] Al-Amleh B, Lyons K, Swain M. Clinical trials in zirconia: asystematic review. J Oral Rehabil 2010;37:641–52.

[7] Sailer I, Feher A, Filser F, Gauckler LJ, Luthy H, Hammerle CH.Five-year clinical results of zirconia frameworks for posteriorfixed partial dentures. Int J Prosthodont 2007;20:383–8.

[8] Zarone F, Russo S, Sorrentino R. Fromporcelain-fused-to-metal to zirconia: clinical andexperimental considerations. Dent Mater 2011;27:83–96.

[9] Guazzato M, Walton TR, Franklin W, Davis G, Bohl C,Klineberg I. Influence of thickness and cooling rate ondevelopment of spontaneous cracks in porcelain/zirconiastructures. Aust Dent J 2010;55:306–10.

[10] Rues S, Kroger E, Muller D, Schmitter M. Effect of firingprotocols on cohesive failure of all-ceramic crowns. J Dent2010;38:987–94.

[11] Asaoka K, Tesk JA. Transient and residual stress in aporcelain-metal strip. J Dent Res 1990;69:463–9.

[12] Swain MV. Unstable cracking (chipping) of veneeringporcelain on all-ceramic dental crowns and fixed partialdentures. Acta Biomater 2009;5:1668–77.

[13] Bonfante EA, da Silva NR, Coelho PG, Bayardo-Gonzalez DE,Thompson VP, Bonfante G. Effect of framework design oncrown failure. Eur J Oral Sci 2009;117:194–9.

[14] Marchack BW, Futatsuki Y, Marchack CB, White SN.Customization of milled zirconia copings for all-ceramiccrowns: a clinical report. J Prosthet Dent 2008;99:169–73.

[15] Tinschert J, Schulze KA, Natt G, Latzke P, Heussen N,Spiekermann H. Clinical behavior of zirconia-based fixedpartial dentures made of DC-Zirkon: 3-year results. Int JProsthodont 2008;21:217–22.

[16] Choi JE, Waddell JN, Swain MV. Pressed ceramics ontozirconia, Part 2: Indentation fracture and influence ofcooling rate on residual stresses. Dent Mater 2011;27:1111–8.

[17] Gostemeyer G, Jendras M, Dittmer MP, Bach FW, Stiesch M,Kohorst P. Influence of cooling rate on zirconia/veneerinterfacial adhesion. Acta Biomater 2010;6:4532–8.

[18] Fischer J, Stawarczyk B, Tomic M, Strub JR, Hammerle CH.Effect of thermal misfit between different veneeringceramics and zirconia frameworks on in vitro fracture loadof single crowns. Dent Mater J 2007;26:766–72.

[19] Fairhurst CW, Hashinger DT, Twiggs SW. Glass transitiontemperatures of dental porcelain. J Dent Res 1981;60:995–8.

[20] Baldassarri M, Stappert CF, Wolff MS, Thompson VP, ZhangY. Residual stresses in porcelain-veneered zirconiaprostheses. Dent Mater 2012;28:873–9.

[21] O’Brien WJ. Dental materials and their selection. HanoverPark, IL: Quintessence Publishing Co, Inc.; 2008.

[22] Guazzato M, Albakry M, Ringer SP, Swain MV. Strength,fracture toughness and microstructure of a selection ofall-ceramic materials, Part II. Zirconia-based dentalceramics. Dent Mater 2004;20:449–56.

[23] Tholey MJ, Swain MV, Thiel N. Thermal gradients andresidual stresses in veneered Y-TZP frameworks. Dent Mater2011;27:1102–10.

[24] Komine F, Saito A, Kobayashi K, Koizuka M, Koizumi H,Matsumura H. Effect of cooling rate on shear bond strengthof veneering porcelain to a zirconia ceramic material. J OralSci 2010;52:647–52.

[25] Tan JP, Sederstrom D, Polansky JR, McLaren EA, White SN.The use of slow heating and slow cooling regimens tostrengthen porcelain fused to zirconia. J Prosthet Dent2012;107:163–9.

[26] Mainjot AK, Schajer GS, Vanheusden AJ, Sadoun MJ.Influence of cooling rate on residual stress profile inveneering ceramic: measurement by hole-drilling. Dent

Mater 2011;27:906–14.

[27] Ivoclar. Programat® EP 3000 – program table; 2009 http://icubusivoclarvivadentcom/Secure30/media area/ivoclar/C/

1 ( 2 0 1 5 ) 77–87

DATA%5Cclient%5Civoclar%5Cmedia%5Cdownloads%5C/Media 611442002pdf/true/application%5Cpdf/EP3000progtab cf enpdf

[28] Ivoclar. Programat® EP 3000 – program table; 2010http://wwwivoclarvivadentcom/zoolu-website/media/document/4611/Programat+-+Firing+program+tables

[29] Guess PC, Bonfante EA, Silva NR, Coelho PG, Thompson VP.Effect of core design and veneering technique on damageand reliability of Y-TZP-supported crowns. Dent Mater2013;29:307–16.

[30] Scherrer SS, Quinn JB, Quinn GD, Wiskott HW. Fractographicceramic failure analysis using the replica technique. DentMater 2007;23:1397–404.

[31] McLean JW. Preparation an the design of metalsubstructures. In: Quintessence, editor. Dental ceramics.Hanover Park: Quintessence; 1983.

[32] Molin MK, Karlsson SL. Five-year clinical prospectiveevaluation of zirconia-based Denzir 3-unit FPDs. Int JProsthodont 2008;21:223–7.

[33] Schmitter M, Mussotter K, Rammelsberg P, Stober T,Ohlmann B, Gabbert O. Clinical performance of extendedzirconia frameworks for fixed dental prostheses: two-yearresults. J Oral Rehabil 2009;36:610–5.

[34] Huang M, Thompson VP, Rekow ED, Soboyejo WO. Modelingof water absorption induced cracks in resin-basedcomposite supported ceramic layer structures. J BiomedMater Res B: Appl Biomater 2008;84:124–30.

[35] Okeson J. Alignment and occlusion of the dentition.Management of temporomandibular disorders andocclusion. 7th ed. Mosby; 2012. p. 58–80.

[36] Abernethy RB. In: Abernethy RB, editor. The new Weibullhandbook fifth edition, reliability and statistical analysis forpredicting life, safety, supportability, risk, cost and warrantyclaims. fifth ed. 2007.

[37] Nicholls D, Lein P. Weibayes testing: what is the impact ifassumed beta is incorrect? Reliability and maintainabilitysymposium, 2009 RAMS 2009 Annual 2009. 2009. p. 37–42.

[38] Denry I, Kelly JR. State of the art of zirconia for dentalapplications. Dent Mater 2008;24:299–307.

[39] Guess PC, Zhang Y, Thompson VP. Effect of veneeringtechniques on damage and reliability of Y-TZP trilayers. Eur JEsthet Dent 2009;4:262–76.

[40] Rekow D, Zhang Y, Thompson V. Can material propertiespredict survival of all-ceramic posterior crowns? CompendContin Educ Dent 2007;28:362–8, quiz 69, 86.

[41] Tholey MJ, Berthold C, Swain MV, Thiel N. XRD2micro-diffraction analysis of the interface between Y-TZPand veneering porcelain: role of application methods. DentMater 2010;26:545–52.

[42] Tholey MJ, Swain MV, Thiel N. SEM observations of porcelainY-TZP interface. Dent Mater 2009;25:857–62.

[43] Koenig V, Vanheusden AJ, Le Goff SO, Mainjot AK. Clinicalrisk factors related to failures with zirconia-basedrestorations: an up to 9-year retrospective study. J Dent2013;41:1164–74.

[44] Belli R, Frankenberger R, Appelt A, Schmitt J, Baratieri LN,Greil P, et al. Thermal-induced residual stresses affect thelifetime of zirconia-veneer crowns. Dent Mater2013;29:181–90.

[45] Coelho PG, Bonfante EA, Silva NR, Rekow ED, Thompson VP.Laboratory simulation of Y-TZP all-ceramic crown clinicalfailures. J Dent Res 2009;88:382–6.

[46] Mainjot AK, Schajer GS, Vanheusden AJ, Sadoun MJ. Residualstress measurement in veneering ceramic by hole-drilling.Dent Mater 2011;27:439–44.

[47] Quinn GD. Fractography of ceramics and glasses.Washington, DC: National Institute of Standards andTechnology; 2007.

s 3 1

design in restorative dentistry: historical review and clinical

d e n t a l m a t e r i a l

[48] Kim JH, Kim JW, Myoung SW, Pines M, Zhang Y. Damagemaps for layered ceramics under simulated mastication. JDent Res 2008;87:671–5.

[49] Rekow D, Thompson V. Near-surface damage-a persistentproblem in crowns obtained by computer-aided design andmanufacturing. Proc Inst Mech Eng H: J Eng Med2005;219:233–43.

[50] Rekow D, Thompson VP. Engineering long term clinicalsuccess of advanced ceramic prostheses. J Mater Sci Mater

Med 2007;18:47–56.

[51] Wiskott HW, Nicholls JI, Belser UC. Stress fatigue: basicprinciples and prosthodontic implications. Int J Prosthodont1995;8:105–16.

( 2 0 1 5 ) 77–87 87

[52] Bhowmick S, Melendez-Martinez JJ, Hermann I, Zhang Y,Lawn BR. Role of indenter material and size in veneer failureof brittle layer structures. J Biomed Mater Res B: ApplBiomater 2007;82:253–9.

[53] Zhang Y, Chai H, Lee JJ, Lawn BR. Chipping resistance ofgraded zirconia ceramics for dental crowns. J Dent Res2012;91:311–5.

[54] Wiskott HW, Belser UC. A rationale for a simplified occlusal

guidelines. J Prosthet Dent 1995;73:169–83.[55] Myers GE, Anderson Jr JR. Nature of contacts in centric

occlusion in 32 adults. J Dent Res 1971;50:7–13.