SCIENTIFIC ARTICLEAustralian Dental Journal 201 0; 55: 306-310

doi: 10.1111/j.1834-7819.2010.01239.x

Influence of thickness and cooling rate on development ofspontaneous cracks in porcelain/zirconia structures

M Guazzato;~ TR Walton, * W Franklin, t G Davis,t C Bohl,t I Klineberg*

.Discipline of Prosthodontics, Faculty of Dentistry, Westmead Centre for Oral Hea/th, Westmead Hospital, The University of S),dney, New SouthWales.tPrivate practice, Sydney, New Soutb Wales.

ABSTRACT

Background: Clinical studies reporting the outcome of zirconia-based restorations indicate that the strength of the zirconiaframeworks is sufficient to withstand masticatory forces. However, a significant incidence of cohesive fracture of theveneering porcelain has been reported. The aim of this study was to investigate spontaneous crack development (chipping,rupture) in a range of porcelains veneered to a zirconia core as a result of thermal stresses induced by changes in thicknessand cooling rate. The hypothesis tested was that crack incidence would increase with increased veneer thickness and fastercooling rates.Methods: Zirconia spheres (diameter 7.8 mm) were veneered with 1.5 gm (thickness ratio 1:2) and 2.5 gm (thickness ratio1:1) of five nominally compatible commercially available porcelains. The manufacturers' firing cycles and a rapid coolingfiring cycle were followed.Results: Multiple regression analysis showed positive associations between the occurrence of cracks and the three covariates(materials, thickness and cooling rate). The incidence of cracks and rupture of the veneering porcelain increased with a fastercooling rate and increased thickness of the specimens in three porcelain-zirconia combinations.Conclusions: Crack incidence increased with increased porcelain veneer thickness and faster cooling rates in nominallycompatible porcelain/zirconia systems in the geometrically configured specimens tested.

Keywords: Zirconia, porcelain, thermal compatibility, residual stresses,crack.

Abbl'eviations and acronyms: CAD/CAM =computer aided designicomputer aided milling; CTE =coefficient of thermal expansion;MCRs =metal ceramic restorations; Y-PSZ =yttria-partially stabilized zirconia.

(Accepted for publication 19 November 2009.)

INTRODUCTION

The clinical applications of ceramic systems such asheat-pressed glass-ceramics, glass-infiltrated ceramicsand fully sintered alumina have been limited by themechanical properties of these materials.1-3 The intro-duction of computer aided design/computer aidedmilling (CAD/CAM) technology into the dental markethas facilitated the use of yttria-partially stabilizedzirconia (Y-PSZ), a strong and tough ceramic.4-7

The few clinical studies reporting the outcome ofY-PSZ restorations appear to indicate that with appro-priate design, the strength of Y-PSZ frameworks issufficient to withstand masticatory forces. However, asignificant incidence of cohesivefracture in the veneeringporcelain has also been observed.8-10 The fracturemechanism of the porcelain layer (cohesive rather thanadhesive) observed in these clinical studies indicates that

306

a lack of adequate bond strength between porcelain andY-PSZ is not the cause for this observed chipping.8-10Other factors such as the design of the Y-PSZframeworks,the mechanical properties of the veneering porcelain or amismatch of the coefficient of thermal expansion (GTE)between porcelain and Y-PSZ have been implicated.

The compatibility between veneering porcelain andmetal substructures has also been identified as a problemfor metal ceramic restorations (MCRs). Walton andO'Brien 11investigated the effect of the mismatch of theGTE and the geometry on thermally induced stressfailures in metal-ceramic discs and spheres. In additionto residual stresses caused by the mismatch of the GTE,significantresidual stresses were generated by differentialcooling rates (thermal gradient) of the materials, partic-ularly evident in thicker and sphere-shaped specimens.

More recently Swain 12 provided a theoretical anal-ysis of the key processing parameters that may

@ 2010 Australian Dental Association

contribute to the development of residual stresses andspontaneous cracking in a range of different veneeredbilayered materials. The analysis identified the impor-tance of cooling rate, thickness and mismatch of theerE in the development of residual stresses. Residualstresses within the porcelain layer may promote thepropagation of a crack caused by contact damage ofthe in situ restoration leading to chipping of theveneering layer. Swain concluded that thick veneerlayers on frameworks with low thermal diffusivity,such as Y-PSZ, are prone to generate fiigh tensilestresses within the porcelain layer. Such stresses mayresult in unstable cracking or chipping. To date thereare no studies investigating the influence of coolingrate and thickness on the development of cracks in arange of commercially available porcelains veneered toY-PSZcores.

The aim of this study was to assess spontaneouscrack development in five nomina\1y compatible porce-lains veneered to Y-PSZ cores as a result of thermalstresses induced by geometry of the specimens, thick-ness and cooling rate. The hypothesis tested was thatcrack incidence would increase with increased veneerthickness and faster cooling rates.

MATERIALSAND METHODS

Two hundred and "twenty Y-PSZ ceramic spheresmeasuring 7.8 mm in diameter were obtained fromone company (Ekton-Straumann, Basel, Switzerland)and veneered with five nomina\1y compatible commer-cially available porcelains (Table 1). Groups of 10specimens were prepared as fo\1ows: group 1.5A(1.5 gm of porcelain applied to the spheres and firedaccording to the manufacturer's recommendations -normal cooling); group 1.5B (1.5 gm of porcelainapplied to the spheres and rapidly cooled by openingthe furnace immediately after the recommended hold-ing time and moving the specimens to the furnace bench- rapid cooling); group 2.5A (2.5 gm of porcelainapplied to the spheres and fired according to themanufacturer's recommendations); and group 2.SB(2.5 gm applied to the spheres and rapidly cooled aspreviously described). Details of the firing conditionsare provided in Table 2.

Table 1. Veneering porcelains

Compatibility of porcelain/zirconia structures

Subsequent to the preparation of the specimens, onemanufacturer changed the recommended firing sche-dule of its porcelain (Vita Zahnfabrik - VM9); the testsfor group I.SA and 2.5A were repeated according tothe new firing schedule (Tables 1 and 2; Fig 1).

Specimens of each group were equally prepared bythree experienced dental technicians to minimize the riskof bias and errors. The specimens prepared with 1.5 gmof porcelain had a mean final diameter of 11.5 mm,

corresponding to a ratio Y-PSZlporcelain of 2:1. The 7 22.5 gm specimenshad a mean final diameter of14.8 mm . ..(t~ckne~~ r~ti.Q1:1).Specimenswere wellcondensed anddried before initiating the sintering process. A thin washlayer (first dentine firing) followed by a dentine layer(second dentine firing) was initially applied to all thespecimens according to the manufacturer>sinstructions(Table 2). The rapid cooling cyclewas applied only after'the second dentine firing of the B groups. After firing,"each specimen was visually examined under magnifica-tion (xlO) and UV light to verify the presence of cracks.

Multiple regression analysis was initially conductedto test the significance of the influence of thickness,cooling and material on the occurrence of crack events.The analysis was performed using a logistic regressionmodel where all the independent variables (covariates)were included to test independent associations.Reduced models were also considered, where associa-tions of the materials, cooling rate and thickness weretested against both a fixed cooling rate and a fixedthickness. Finally, the Kruskal-Wallis test was per-formed to re-evaluate the consistency of the resultsobtained from the logistic regression model. Thestatistical analysis was conducted using the softwarePackage R (Revolution Computing).13

The VM9-N groups were excluded from the statis-tical analysis because the porcelain in these groups wasfired with a slower cooling rate, not comparable to thatof the other groups.

RESULTS

The multiple regression analysis showed independentpositive associations between the occurrence of cracksand the three covariates (materials, thickness andcooling rate). The statistical analysis showed that

.VM9 fired with the firing schedule as recommended before March 2009.tvM9 fired with the firing schedule as recommended after March 2009.

~ 2010 Australian Dental Association 307

Abbreviation Porcelain Lot number Manufacturer

NR Nobel Rondo 0608 Nobel Biocare- Zurich, SwirzerlandICE Ice KA80460A Zirkonzahn - Gais, ItalyZR Cerabien ZR 016208 Noritake -Higashiyama, Japan21 ZI-F 0496 Creation - Meiningen, AustriaVM9 - 0 VM9' 27710 VITA Zahnfabrik -Bad Siickingen,GeLmanyVM9-N VM9t 27710 vcr A Zahnfabrik- Bad Siickingen, Gennany

For each group the initial abbreviation corresponds to the veneering porceliiin (e.g., NR corresponds to Nobel Rondo) as illustrated by Table 1; thenumber corresponds to the quantity of porcelain used for the specimens and therefore the relative thickness (e.g., 1.5 corresponds to 1.5 grams); theletter A or B correspond to the cooling rate (A =according to manufacturer; B " rapid cooling).

Fig 1. Specimens used for the test. (From left to right) zirconia sphere;bilayered porcelain/zirconia specimen made with 1.5 grams of

porcelain: bilayered porcelain/zirconia specimen made with2.5 grams of porcelain; one of fractured specimens.

increasing thickness was positively associated withcrack events (p =0.000049). Rapid cooling (B groups,Table 2) was also significantly associated withincreased crack events (p =0.0018). Of the five com-binations of porcelain/Y-PSZ ceramic used in thisstudy, VM9-0 (p = 0.00014), ICE (p = 0.00000025)and ZI (p = 0.01) were significantly associated with thespontaneous development of cracks upon firing.

A statistical analysis performed on a reduced modelwith thickness (2.5 groups, Table 2) as the fixedparameter showed results consistent with the fullmodel. In this instance, the cooling rate (p = 0.001),ICE (p = 0.0001), VM9-0 (p = 0.0012) and ZI(p ;: 0.0056) had a positive association with thevariation in thickness. Similar results were observed

30B

Table 3. Crack development in each group (specimenscracked/sample size)

Material Group 1.5A Group 1.5B Group 2.5A Group 2.5B

NRICEZRZIVM9-0VM9 -._N

0/100/100/100/102/100/10

0/100/100/100/103/10N/A

0/100/100/102/104/100/10

0/10

2/10 t>. P -[,<;;0/10.."... ~8/10/

8/10N/A

when the analysis was performed with the cooling rate(B groups, Table 2) as the fixed parameter. In thisinstance, a positive association was reported forthickness (p;: 0.00016), ICE (p = 0.00002), VM9-0(p = 0.0017) and ZI (p = 0.018).

The Kruskal-Wallis rank sum test as non-parametricanalysis confirmed the results obtained with multipleregression analysis (p ;: 0.0001).

The number of cracked or ruptured specimens isreported in Table 3. The cracks consistently ranvirtually parallel to the surface and in some samplespropagated thrQughout the porcelain layer causingrupture. The rupture was consistently cohesive withoutexposure of the Y-PSZ core (Fig 1).

DISCUSSION

A mathematical analysis by Swain12 indicated themismatch of the CTE; the thickness and the coolingrate play an important role in the development ofresidual stresses within porcelain/Y-PSZ prostheses.Such stresses may be responsible for chipping of the

@ 2010 Australian Dental Association

M Guazzato et a/.

Table 2. Firing conditions

Group Layer Heating Firing Extended Normal Rapidrate temp. cooling cooling cooling

NR 1.5Aand NR 2.5A 1st dentine 45°C 910 DC - Yes2nd dentine 45 °C 900 DC - Yes

NR 1.5Band NR 2.5B 1stdentine 45°C 910 DC - - Yes1stdentine 45 °C 900 DC - - Yes

ICE 1.5A and ICE 2.5A 1st dentine 45°C 920 DC - Yes2nd dentine 45 °C 820 DC - Yes

ICE 1.5Band ICE2.5B 1stdentine 45°C 920 DC - - Yes1stdentine 45 °C 820 DC - - Yes

ZR 1.5Aand ZR 2.5A 1st dentine 45°C 930 DC - Yes

2nd dentine 45 °C 930 DC - Yes

ZR 1.5B and ZR 2.5B 1st dentine 45°C 930 DC - - Yes

1st dentine 45 °C 930 DC - - YesZI 1.5Aand ZI2.5A 1st dentine 45°C 810 DC - Yes

2nd dentine 45 °C 800 DC - Yes

ZI 1.5B and ZI 2.5B 1st dentine 45°C 810 DC - - Yes1stdentine 45 °C 800 DC - - Yes

VM9-0 1.5Aand VM9-0 2.5A 1st dentine 55°C 950 DC - Yes2nd dentine 55 °C 910 DC - Yes

VM9-0 1.5 Band VM9-0 2.5B 1stdentine 55°C 950 DC - - Yes

1st dentine '55°C 910 DC - - YesVM9-N 1.5Aand VM9-N2.5A 1stdentine 55°C 950 DC 600 DC

2nd dentine 55 °C 800 DC 600 DC

porcelain reported in clinical studies.8-10 The presentstudy showed that for some of the groups, spontaneouscracking and cohesiv~racture of the porcelainincreased with an increase in the porcelain thicknessand faster cooling rates. These results appear to supportthe concept that mismatch of the erE, cooling rate andthickness may play an important role in explainingchipping observed in porcelain/Y-PSZ prostheses.

The spherical geometric form of the specimens was. utilized as it has been shown to induce maximum

thermal stresses in bilayered structures and r~resentsthe extreme form that can occur clinically.H,l ,15

DeHoff et al.l5 showed that bilayer ceramic cylindersand spheres provide confirmation of thermal incom-patibility stress that was predicted by finite elementanalysis. They veneered cylindrical and spherical pres-sable ceramic cores with four thermally incompatible.dental ceramics (with markedly different CTE) and fourthermally compatible dental porcelains (with CTE in a .nominally compatible range). When spheres were usedas the core material, 100% of the thermally incompat-ible specimens failed, whereas only 4% of the thermallycompatible specimens developed cracks after firing.Finite element analysis confirmed that failure wasconsistently linked to high residual stresses, whereaslow residual stresses were measured for those specimensthat did not fail. The results of this study indicate thatthe use of either cylindrical or spherical specimens is asimple and reliable method that can be used to identifyhighly incompatible ceramic combinations.

All porcelains tested in the present study have,according to the manufacturers, a CTE that rendersthem thermally compatible with that of Y.PSZ mate-rials. However, in some groups the percentage of failedspecimens was greater than that reported by DeHoffet al.15for nominally compatible ceramic systems, evenwhen the manufacturer's firing schedule was followed.Multiple regression analysis showed that cooling rate,thickness and material were associated with the devel-opment of spontaneous cracks. Surprisingly, even whenthe manufacturer's firing schedule was followed, crackswere observed in two of the six systems (VM9-0 1.5Aand 2.5A, and ZI 2.5A).

Manufacturers tend to produce veneering porcelainwith a erE slightly lower than the CTE of the

1 framework so that compressive stresses are develo edON "jn the porce ain sur ace. One can assume that the

dIfference in results between the six systems tested isnot only affected by the thickness and the cooling rate,but also by minimal differences in mismatch of CTEamong the porcelain/Y-PSZ systems given the samegeometry and, therefore, thermal diffusivity of thespecimens. To date, the range of CTE that is defined ascompatible in porcelain/Y-PSZ structures has not beeninvestigated. However, it is likely that the range ofcompatible CTE in porcelain/Y-PSZ structure is con-@ 2010 Australian Dental Association

Compatibility of porcelain/zirconia structures

siderably narrower than that measured for MCR.11,14In the latter, some of tile developing residual thermalstress is relieved by creep deformation.l6 Y-PSZ is ahigh fusing brittle ceramic material not able to relievethermal stress by deformation under the conditionstested here, or in clinical situations, and therefore aminimal mismatch of the CTE may count for thedevelopment of significant residual stress even innominally compatible systems. This hypothesis shouldbe confirmed by a quantitative analysis of the residualstresses in the systems investigated.

The importance of the cooling cycle was especiallydemonstrated by the results of the VM9 specimens.Cracks were observed in the VM9 specimens as thethickness was increased and, more importantly, whenthe cooling rate was faster (the originally recommendedcooling cycle). However, no cracks were seen when theupdated firing recommendations involving slow cool-ing of the specimens to 600°C were used (comparedto 800 °C previously recommended). These findingssupport the assumption that in addition to a potentialmismatch of the CTE of the materials, the thickness ofthe structure and cooling rates may be even moresignificant factors responsible for the development ofcracks and chipping of porcelain/Y-PSZ prostheses.

Tempering-induced residual tensile stress developswithin the porcelain layer when it cools below the glasstransi~ion phase. Above this temperature, stresses arerelieved by plastic deformation. Temperature gradientsthrough the cooling porcelain can result in areas thatare above and below this glass transition phase. Thesetemperature gradients will be affected by the coolingrate, thickness and thermal conductivity of both theporcelain and zirconia core. Because the erE is notlinear with temperature change, small differences inCTE through the porcelain layer and between theporcelain and zirconia are accentuated. The faster thecooling rate, the greater the temperature gradientsthrough the porcelain. The external regions will coolfaster, thus concentrating stresses near the surface.Therefore, the thicker the porcelain, the faster thecooling rate, and the slower the thermal diffusivity(thermal conductivity), then the greater will be the

stress development. ~) .

Swain's mathematical model predicted that with ~ ,.,>9-v 0 \~~small differences in CTE compressive stresses develop 0 ~"" C,Oat the porcelain/Y-PSZ interface, whereas tensile r""'~ ,,'I}!;stresses concentrate at the surface of the porcelain1ayer.' ",U/~J"'" III<i.-Th1Sis consistent with the observatIon of tl~and ~ ~~\ ~rupture of the specimens in this study, since the cracks l'.rJvJ{ "0 U .-appeared to originate within the porcelain layernear the r 1j't.-~r:1 ~ '7 v,J5(surface and when rupture occurred it never extended to ~ \ r ().~\

the interface where presumably the layer of compressive ~"" 7 '~~stresses diverted its propagation (Fig 1). rJ--'

Zirconia has low thermal conductivity compared to ('(~metals, and zero creep at porcelain firing temperatures.

309

M Guazzato et al.

Thus, given the same geometry (shape and thickness),same cooling rate and same CTE differential, higherstresses are likely to develop in Y-PSZ combinationscompared to porcelain/metal combinations.

The role of thickness in the development of unstablecracks is controversial. Someclinical studies support theview that chipping of the porcelain layer can beeliminated by increasing the thickness of Y-PSZ and byreducing the thickness of the porcelain layer.9.17How-ever, according to Swain,12 when materials have poorthermal diffusivity, such as Y-PSZ and porcelain, theeffective thickness for stress development corresponds tothe total thickness of the restoration and thereforechanging the ratio within the same restoration may haveno effect on the development of tensile stresses.

It is important to note that the lack of visible cracksin some porcelains tested does not necessarily corre-spond to the lack of residual stresses within thestructures. This would also apply to in situ restorationsand does not guarantee they would be immune fromchipping. Crack development and slow growth may beinitiated by contact forces, facilitated by the residualstress, and exacerbated by the presence of moisture.

This study suggests that the development of ther-mally induced residual stress affected by several factorsincluding overall thickness, shape and cooling rate maycontribute to the chipping of the porcelain observedclinically in restorations made from nominally ther-mally compatible porcelain/Y-PSZ systems. Furtherresearch is required to quantify the contribution of eachfactor. This would enable protocols to be implementedto minimize the development of thermal residualstresses and in turn the number of clinical complica-tions experienced with porcelain veneered to Y-PSZprostheses.

CONCLUSIONS

Crack incidence increased with increased porcelainveneer thickness and faster cooling rates in nominallycompatible porcelain/zirconia systems in the geometri-cally configured specimens tested.

ACKNOWLEDGEMENTS

The authors are grateful to Dr Fabio Luciani for thestatistical analysis, to Ekton-Straumann for the fabri-cation of the zirconia spheres and to Vita Zahnfabrik,Creation, Noritake-Alphabond Australia, Zirkonzahnand Nobel Biocare for supplying the porcelain.

REFERENCES

1. Conrad HJ. Seog WJ, Pesun IJ. Current c;eramic materials andsystems with clinical recommendations: a systematic review. JProsthet Dent 2007;98:389-404.

310

2. Sailer I, Pjetursson BE, Zwahlen M., Hammerle CHF. A system-atic review of rhe survival and complication rates of all-ceramicand metal-ceramic reconstructions after an observation period ofat least 3 years. Part II: fixed dental prostheses. Clin OralImplants Res 2007;18(Suppl 3):86-96.

3. Pjetutsson BE, Sailer I, Zwahlen M, Hammerle CHF. A system-atic review of the survival and complication rates of all-ceramicand metal-ceramic reconstructions after an observation period ofat least 3 years. Part I: single crowns. Clin Oral Implants Res2007;18(Suppl 3):73-85.

4. Denry I, Kelly R. State of the art of zirconia for dental applica-tions. Dent Mater 2008;24:299-307.

5. Kelly R, Denry I. Stabilized zirconia as a structural ceramic: anoverview. Dent Mater 2008;24:289-298.

6. Guazzato M, Albakry M, Swain MY, Ringer SP. Strength, frac-ture roughness and microstructure of a selection of all-ceramicmaterials. Part I: pressable and alumina glass-infiltrated ceramics.Dent Mater 2004;20:441-448.

7. Guazzato M, Albakry M, Swain MV, Ringer SP. Strength, frac.ture toughness and microstructure of a selection of all-ceramicmaterials. Part U: zirconia-based dental ceramics. Dent Mater2004;20:449-456.

8. SailerI, Feher A, Filser F, Gauckler LJ, LUthy H, Hammerle CHF.Five-year clinical results of zirconia frameworks for posteriorfixed partial dentures. Int J Prosthodom 2007;20:383-388.

9. Tinscherr J, Schulze KA, Nan G, Laczke P, Heussen N, Spieker.mann H. Clinical behavior of zirconia-based fixed partial den-tures made of DC-Zirkon: 3-year results. Int ] Prosthodonr2008;21:217-222.

10. Raigrodski AJ, Chiche ~J, Potiket N, et al. The efficacy of pas.terior three-unit zirconium-oxide-based ceramic fixed partialdental prostheses: a prospective clinical pilot study. J ProsthetDent 2006;96:237-244.

11. Walton TR, O'Brien WJ. Thermal stress failure of porcelainbonded to a palladium-silver alloy.] Dent Res 1985;64:476-480.

12. Swain MV. Unstable cracking (chipping) of veneering porcelainon all-ceramic dental crowns and fixed partial dentures. ActaBiomater 2009;5:1668-1677.

13. Team ROC. R: A Language and Environment for StatisticalComputing. Vienna: R Foundation for Statistical Computing,2009.

14. Nielsen JP. Tucillo JJ. Calculation of interfacial stress in dentalporcelain bonded to gold alloy substrate. J Dent Res1971;51:1043-1047.

15. DeHoff P, Barretr AA, Lee RB, Anusavice K]. Thermal compat-ibility of dental ceramic systems using cylindrical and sphericalgeometries. Dent Mater 2008;24:744-752.

16. Bertolotti RL. Calculation of interfacial stress on porcelain-fusedto metal systems. J Dent Res 1980;59:1972-1977.

17. Marchack BW, Futatsuki Y, Marchack CB, White SN. Custom.ization of milled zirconia copings for all-ceramic crowns: a clin-ical report. J Prosthet Dent 2008;99:169-173.

Address faT correspondence:Dr Massimiliano Guazzato

Discipline of ProsthodonticsFaculty of Dentistry

Westmead Centre for Oral HealthWestmead Hospital

The University of SydneySydney NSW 2006

Email: mguazzato@yahoo.com.au

@2010 Australian DentalAssociation