http://jdr.sagepub.com/Journal of Dental Research

http://jdr.sagepub.com/content/70/6/1009The online version of this article can be found at:

 DOI: 10.1177/00220345910700060201

1991 70: 1009J DENT RESK.J. Anusavice and B. Hojjatie

Effect of Thermal Tempering on Strength and Crack Propagation Behavior of Feldspathic Porcelains  

Published by:

http://www.sagepublications.com

On behalf of: 

International and American Associations for Dental Research

can be found at:Journal of Dental ResearchAdditional services and information for    

  http://jdr.sagepub.com/cgi/alertsEmail Alerts:

 

http://jdr.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

http://jdr.sagepub.com/content/70/6/1009.refs.htmlCitations:  

What is This? 

- Jun 1, 1991Version of Record >>

at MCGILL UNIVERSITY LIBRARY on September 26, 2013 For personal use only. No other uses without permission.jdr.sagepub.comDownloaded from at MCGILL UNIVERSITY LIBRARY on September 26, 2013 For personal use only. No other uses without permission.jdr.sagepub.comDownloaded from at MCGILL UNIVERSITY LIBRARY on September 26, 2013 For personal use only. No other uses without permission.jdr.sagepub.comDownloaded from at MCGILL UNIVERSITY LIBRARY on September 26, 2013 For personal use only. No other uses without permission.jdr.sagepub.comDownloaded from at MCGILL UNIVERSITY LIBRARY on September 26, 2013 For personal use only. No other uses without permission.jdr.sagepub.comDownloaded from at MCGILL UNIVERSITY LIBRARY on September 26, 2013 For personal use only. No other uses without permission.jdr.sagepub.comDownloaded from

Effect of Thermal Tempering on Strengthand Crack Propagation Behavior of Feldspathic Porcelains

K.J. ANUSAVICE and B. HOJJATIE

Department of Dental Biomaterials, College of Dentistry, University of Florida, Gainesville, Florida 32610-0446

The objective of this study was to test the hypothesis thattempering stress can retard the growth of surface cracks inlayered porcelain discs with variable levels of contraction mis-match. Porcelain discs, 16 mm in diameter and 2 mm thick,were prepared with a 0.5-mm-thick layer of opaque porcelain(0) and a 1.5-mm-thick layer of body porcelain (B). The ma-terials were selected to produce contraction coefficient differ-ences, ao-%, of + 3.2, + 0.7, -0.9, and -1.5 ppm/0C.Body porcelain discs with a thickness of 2 mm were used asthe thermally compatible control specimens (Aa= 0). The discswere fired to the maturing temperature of body porcelain (9820C)and were then subjected to three cooling procedures: slow cooling(SC) in a furnace, fast cooling (FC) in air, and tempering (T)by blasting the surface of the body porcelain with compressedand dried air for 90 s. The dimensions of cracks induced by aVickers microhardness indenter under a load of 4.9 N weremeasured at baseline and six months after indentation at 80points along diametral lines within the surface of body por-celain. In addition, biaxial flexure tests were performed todetermine the influence of mismatch and tempering on flexurestrength. The results of ANOVA indicate that crack dimen-sions were influenced significantly by the interaction of cool-ing rate and contraction mismatch (p<0.0001). Multiple contrastanalysis by the Tukey's HSD Test indicated that the cracklengths of tempered specimens at baseline and six months weresignificantly smaller (p <0.05) than the corresponding valuesfor the FC and SC specimens. For tempered specimens withcontraction differences of + 3.2, 0, and - 1.5 ppm/C, themean crack lengths during the six-month period increased by10.8%, 8.3%, and 9.7%, respectively, compared with in-creases of 13.8%, 20.1%, and 15.9%, respectively, for the SCspecimens. Tempering treatment of the compatible discs(Aa= 0) resulted in the highest mean flexure-strength valueof 116.2 MPa. This value was 2.6 times greater than the corre-sponding value for the slow-cooled specimens. These resultsindicate that tempering by forced convective cooling in airsignificantly strengthened bilayered discs and reduced the in-itial size of induced surface cracks. However, tempering stresswas less effective in reducing the propagation rate of inducedcracks.

J Dent Res 70(6):1009-1013, June, 1991

Introduction.

Previous methods for strengthening dental ceramics have beenbased on (1) compressive stress induced in porcelain throughthe use of a metal substrate with a contraction coefficient greaterthan that of its porcelain veneer, (2) a glaze layer formed onthe external surface to minimize the influence of flaws, (3) aglaze layer fired with a contraction coefficient lower than thatof the adjacent ceramic to produce a state of compressive stress

within the porcelain surface (Creyke, 1968), and (4) ion-ex-change strengthening that induces compressive stress withinthe outer surface because of the substitution of larger for smalleralkali cations. The glazing technique is now used routinely fordental ceramics, although the stress states in glazed surfaceshave not been well-characterized. Thermal tempering, whichis a process of rapidly cooling a glass or ceramic to induce astate of compression in the surface, is an attractive techniquefor strengthening dental ceramics because the technique canbe easily performed in commercial dental laboratories and thematerial surfaces are not degraded as a result of the temperingtreatment. However, viscous flow may occur if the initial tem-pering temperature is too high (Gardon, 1980).

Structural failures of metal-ceramic restorations still occurdespite the known success of specific products and reliabletechniques. The causes of these failures are often multifactorialin nature and can be associated with non-uniform or largeporcelain/metal thicknesses ratios, thermal incompatibility stressin metal-ceramic and ceramic-ceramic (layered) systems, thepresence of critical structural flaws, and non-standardizedprocessing techniques. Chemical and thermal strengtheningtechniques have been shown to be effective in enhancing theperformance and longevity of some industrial glasses and ce-ramics. Doremus (1973) indicated that glasses can be strength-ened by a factor of six for quench-hardening, 10 for ionexchange, and 30 for etching. Ion-exchange strengthening offeldspathic porcelains is a chemical process whereby large po-tassium ions are substituted in the surfaces of glasses and ce-ramics for smaller sodium ions, thereby creating a beneficialstate of compressive stress within the near-surface region. Dunnet al. (1977) placed porcelain bars in a potassium nitrate (KN03)molten salt bath for a period of four h at a temperature of400'C and reported an increase in flexure strength of the speci-mens by a factor of two. Jones (1983) reported that immersingglazed specimens of air-fired feldspathic porcelain bars into apotassium bromide (KBr) salt bath at a temperature of 830'Cfor a period of 12 min would result in an increase of themodulus of rupture from 57.1 MPa (untreated) to 80.4 MPa.However, strengthening by ion exchange in a molten salt bathhas not been used routinely for dental applications because ofthe modest improvement in strength, compared with the dis-advantages that include the hazards associated with molten saltand the possible need for long exposure times in order for asufficient depth of compressive stress to be achieved.

The effectiveness of the tempering technique depends onfactors such as thermal contraction mismatch, initial coolingtemperature, heat transfer coefficients, specimen dimensions,and flaw characteristics. Traditionally, residual stresses pro-duced at the surface of a thermally-treated glass have beendetermined by the use of the photo-elastic birefringence tech-nique. However, measurements of these thermally-inducedstresses in ceramics cannot be based on this optical methodbecause of their inherent opacity. Strain gauge measurementscannot be used because the fabrication temperatures are toohigh. DeHoff and Anusavice (1989) developed an analyticalmodel that incorporated linear visco-elasticity and structuralrelaxation effect to calculate transient and residual stresses in

1009

Received for publication June 15, 1990Accepted for publication February 6, 1991This study was supported by NIDR Grant No. DE06672.

1010 ANUSAVICE & HOJJATIE

dental porcelain subjected to tempering. However, the resultswere not compared with any experimental data.

Marshall and Lawn (1977, 1978a, b, and 1979) used themicrohardness indentation technique to create surface cracksin order to analyze the surface stresses in ceramics. Gupta andJubb (1981) used this technique to measure the growth rate ofcracks in the presence of stress and moisture. Anusavice et al.(1989, 1991) have shown that tempering of feldspathic por-celain can significantly reduce the size of cracks induced by amicrohardness indenter. The objective of this study was to testthe hypothesis that tempering stresses can retard the growth ofsurface cracks in layered porcelain discs with variable statesof thermal contraction mismatch.

Materials and methods.Experimental feldspathic porcelains with variable thermal

contraction mismatch were prepared by J.F. Jelenko and Co.(Armonk, NY). We prepared these porcelains as circular discs16 mm in diameter and 2 mm thick. Each disc consisted of a0.5-mm-thick layer of opaque (0) porcelain and a 1.5-mm-thick layer of body (B) porcelain. These thicknesses were con-trolled by grinding the surface of each layer with silicon car-bide abrasives and subsequently measuring the thickness witha micrometer. The thickness of the body (B) porcelain controldisc was 2 mm. Five specimens for each porcelain were pre-pared, also as rectangular bars (5 x 5 x 50.8 mm), and anOrton Dilatometer (Orton Ceramic Foundation, Columbus, OH)was used to measure the thermal contraction coefficient (a)for each bar according to the ASTM standard (DesignationE228-71). The mean values of thermal contraction coefficientsfor each porcelain were determined from contraction curvesfor the specimens cooled from 500'C to 250C at a rate of 3YC/min. Listed in Table 1 are the relative contraction differences(ao-a%) of the four groups of opaque and body porcelain com-binations and the all-body porcelain control.The wet body porcelain was condensed into a cylindrical

brass mold (approximately 3.5 mm in depth) and vibrated witha Vibra II handpiece (J.F. Jelenko and Co., Armonk, NY).Excess moisture was blotted dry with a tissue. The surface ofthe mold was trimmed flush with a straight-edge razor blade,and the opaque porcelain was applied manually on this surface.The specimens were dried in front of the open door of a NeyMark IV Digital furnace (J.M. Ney Company, Bloomfield,CT) at an initial temperature of 6480C for ten min and werethen placed inside the furnace to be fired. The temperature ofthe furnace was raised to a maximum temperature of 9820C ata heating rate of 550C/min. The discs were ground on theopaque- and body-porcelain surfaces with a Buehler polishingwheel (Buehler Ltd., Lake Bluff, IL) and silicon carbide abra-sive, beginning with 120-grit paper and ending with 600-gritpaper. After the layers were polished with 1-pim A1203 abra-sive, the thickness of each layer was measured from the edgeof the disc by use of a micrometer with a tolerance of 0.01

mm. In a second firing cycle, the discs were fired to 982°Cand then subjected to one of the three cooling methods: (1)slow cooling (SC) by termination of power to the furnace, (2)fast cooling (FC) by being quenched in ambient air, and (3)tempering (T) by a compressed air blast directly upon the discimmediately upon its removal from the furnace. A nozzle witha 4-mm diameter (Anusavice et al., 1989) was placed 20 mmabove the disc surface, and dry compressed air was blasted onone side of the disc at a pressure of 0.34 MPa for 90 s.A temperature vs. time profile was determined for each of

the three cooling methods by a chromel-alumel (K type) ther-mocouple that was placed within the porcelain disc (prior tobeing fired) with the tip positioned at a depth of 0.5 mm fromthe center of one of the flat disc surfaces. The thermocouplewas interfaced with an IBM PC-XT microcomputer that re-corded and plotted the thermal history for each of the threecooling methods (SC, FC, and T). Beginning at the final firingtemperature of 982°C, the ambient temperature of 25°C wasreached after approximately four h for slow cooling (SC), 7.5min for fast cooling (FC), and 30 s for tempering (T).A Vickers indenter was applied to the body porcelain sur-

faces at a load of 4.9 N. Each disc was indented along fourdiametral lines, each receiving 20 indentations spaced 300 pmapart (Fig. 1). Thus, each disc received 80 indentations. Mea-surements of the crack length were made in the body porcelainsurface with a measuring microscope at a magnification of20 x . Each reading was made 45 s after indentation (baseline)to minimize errors in measurement of crack length (2c) due tocontinuing crack propagation in the presence of residual in-dentation stress and environmental moisture. Additional mea-surements were made after the specimens were stored in ourlaboratory for six months. The relative humidity of the air inthe laboratory was 55 + 2%, which was measured periodicallyover the six-month period. The mean 2c-dimensions of theradial "half-penny" cracks-which were measured at baselineand six months for each of the five states of contraction mis-match and three cooling techniques -were analyzed by a three-factor analysis of variance (ANOVA). In order for the inter-action effect of time, cooling rate, and contraction differenceto be examined, the Tukey (HSD) multiple-contrast analysiswas used at the 5% level of significance to determine the spe-cific conditions that had a significant effect on the crack length(2c).The surface stress values (MPa) for the tempered (T) speci-

mens were calculated according to relations proposed by Mar-shall and Lawn (1978a). From the five mismatch groups, onlythree groups of porcelains with thermal contraction differencesof + 3.2, 0 (control), and - 1.5 were selected for biaxial flex-ure-strength testing. Six discs, 16 mm in diameter and 2 mmin thickness, were prepared for each mismatch group and eachcooling condition. The discs were subjected to biaxial flexure(piston-on-three-ball) under water at 3TC. Each disc was placedupon three equidistant stainless-steel balls (3.2 mm in diame-ter) that lay upon the perimeter of a circle 10 mm in diameter(Ban and Anusavice, 1990). The discs were loaded at the cen-

TABLE 1THERMAL CONTRACTION CHARACTERISTICS OF PORCELAIN

SYSTEMS USED FOR EXPERIMENTAL DISCS

Opaque Porcelainao (ppm/°C)

14.214.213.5*12.612.0

Body PorcelainaB (ppM/C)

11.013.513.513.513.5

*Represents body porcelain as a control disc.

MismatchaO-(XB (ppm/°C)

+3.2+0.70.0*

-0.9-1.5 af2Co -

Fig. 1-Illustration of indentation sites.

- 20INDETATIONSMINE

_ DNWmmOUATSA(NOT ORAWN TO SCALE)

J Dent Res June 1991

THERMAL TEMPERING OF FELDSPATHIC PORCELAIN

160

140

o 120L.

100

z

wU 80

U 60

40

20+3.2 +0.7 0 -0.9 -1.5

ACI(ppm/'C)Fig. 2-Mean crack length of tempered vs. fast-cooled specimens at

baseline and six months for each level of thermal contraction mismatch.

to

0

h._

I--

zw

-J

4

C.)

+3.2 +0.7 0 -0.9 -1.5Aa(ppm/0C)

Fig. 3-Mean crack length of tempered vs. slow-cooled specimens atbaseline and six months for each level of thermal contraction mismatch.

ter of the top surface by means of a steel piston (with 1.2-mmdiameter flat-ground along the contacting surface) at a loadingrate of 0.5 mm/min until fracture occurred. This mode of flex-ure caused tensile failure to initiate at the center of the lower(tempered) surface of the disc. The failure stress, o, at thecenter of the lower surface was calculated from the followingequations developed by Marshall (1980) for the piston-on-three-ball flexure test described by Wachtman et al. (1972):

A P0c t

= flexure strengthA = (3 / 4rr) [2 (1 + v) ln (a/r0*) + (1 - v) (2a2 - r.*2)/

2b2 + (1 + v)]P = applied load at failurev = Poisson's ratio. In this study, a value of 0.28 was used

for dental porcelain (Farah and Craig, 1975).a = radius of the support circle

For small ro, as in this case: r0* = (1.6 r02 + t2)1/2-0.675t.

r. = radius of the piston at the contacting surfaceb = radius of disc speciment = thickness of the disc specimen

The failure stress data were analyzed by two-factor ANOVAand Tukey's Multiple Range Test to determine whether themean flexure-strength values were significantly influenced bythe cooling conditions and contraction differences.

Results.Based on a three-factor analysis of variance, the influence

of the cooling technique and thermal contraction mismatch atbaseline and at six months and their interaction effect on thecrack length were highly significant (p<O.OOO1). Summarizedin Table 2 are the results of the Tukey's Multiple Contrastanalysis of the mean crack lengths (2c) and the 95% confidenceintervals corresponding to the baseline and six-month data forthe slow-cooled, fast-cooled, and tempered groups for the fivelevels of contraction difference. Based on the Tukey test analy-sis, the mean crack length (pm) for each contraction mismatchgroup was compared for the three cooling conditions at base-line and six months. Groups that are connected by the samehorizontal line were not significantly different from one an-other at the 0.05 level of confidence. The results shown in thisTable indicate that for a given contraction mismatch level, themean crack diameter at baseline for tempered discs (T) wassignificantly smaller (p<0.05) than that of the correspondingvalues for the slow-cooled (SC) and fast-cooled (FC) groups.Except for the control discs (Aa = 0.0), there was no signif-icant difference in crack length at baseline between SC andFC groups. In each porcelain combination, the crack lengthafter six months for the tempered discs (T) was significantlysmaller than that of the SC and FC discs. There was no sig-nificant difference between mean values at six months of SCand FC for contraction mismatch groups of + 3.2, + 0.7, and- 1.5. Shown in Figs. 2 and 3 are the mean crack lengths forthe five levels of mismatch at baseline and six months, corre-

TABLE 2RESULTS OF THE TUKEY TEST ANALYSIS AND 95% CONFIDENCE INTERVALS FOR SLOW-COOLED (SC), FAST-COOLED (FC), AND

TEMPERED (T) SPECIMENS AT EACH LEVEL OF THERMAL CONTRACTION MISMATCH

Mean Crack Length (pm)Aaa SC FC SC FC T

(ppm/0C) 6 Months 6 Months Baseline Baseline 6 Months Baseline+3.2 117.5±3.5 112.0±3.2 103.3 ±2.8 103.6± 2.6 99.0±3.2 89.4±2.4

+0.7 125.0±4.8 126.3 ±5.7 102.1±2.9 101.6±3.3 89.9±2.9 81.2±2.1

0.0 138.1±5.0 115.3±3.5 115.0±3.5 101.9±2.9 81.4±1.8 75.1±1.6-0.9 112.2±3.4 100.5 ±2.9 93.6±3.0 88.6±2.8 91.8±2.3 70.5±1.6

-1.5 106.5±3.2 101.4±2.5 91.9±2.4 91.6±3.0 83.2±2.1 75.8±1.8

Values joined by horizontal lines are not significantly different at p = 0.05.

Vol. 70 No. 6 1011

1012 ANUSAVICE & HOJJATIE

TABLE 3TUKEY'S MULTIPLE CONTRAST ANALYSIS OF MEAN CRACKLENGTH (pm) FOR THE THREE COOLING CONDITIONS AT

BASELINE AND SIX MONTHS AT EACH LEVEL OFCONTRACTUAL MISMATCH

Mean Crack Length (I.m)Aa (ppm/C) SC2 FC2 SC1 FC1 T2 T,

+3.2 117.5 112.0 103.3 103.7 99.0 89.4+0.7 125.0 126.3 102.1 101.6 89.9 81.20.0 138.1 115.3 115.0 101.9 81.4 75.1

-0.9 112.2 100.5 93.6 88.6 91.8 70.5-1.5 106.5 101.4 91.9 91.7 83.2 75.9

1 = Baseline; 2 = Six months.Groups that are joined by a horizontal line are not statistically different

at p = 0.05.

TABLE 4COMPUTED SURFACE STRESSES* IN BODY PORCELAIN AS AFUNCTION OF SLOW COOLING (SC), FAST COOLING (FC), and

TEMPERING (T)

x-aB (ppmPQC) as (MPa) aFC (MPa) cT (MPa)+3.2 +47.5 -7.7 -34.5+0.7 +10.4 -17.0 -64.10.0 0.0 -16.5 -86.7

-0.9 -13.4 -46.3 -109.5-1.5 -22.3 -35.5 -83.5

*DIsc values were determined from finite element analyses.oaF< and aT were determined from the relation developed by Marshall

and Lawn (1979).Negative values of stress represent a state of compression.Positive values of stress represent a state of tension.

sponding to the tempered (T) vs. fast-cooled (FC) groups andtempered (T) vs. slow-cooled (SC) groups, respectively. Nosignificant difference (p>0.05) existed between the crack lengthsof tempered specimens at baseline and six months for contrac-tion mismatch levels of 0 and - 1.5 (Table 3). For each levelof contraction mismatch, the mean crack length at baseline fortempered specimens was significantly smaller than that of thecorresponding values for the slow- and fast-cooled groups.Similarly, the mean crack lengths at six months for temperedspecimens were significantly smaller than the values for slow-cooled and fast-cooled cases.The computed surface stresses in body porcelain for SC,

FC, and T groups were reported previously (Anusavice et al.,1989) and are summarized in Table 4. The residual stress withinthe body porcelain surface of slowly cooled discs was deter-mined by finite element stress analysis and previously mea-sured contraction data (25-500'C) for these experimentalporcelains. For the SC groups, the maximum compressive stress(22.3 MPa) was produced in the disc with the smaller negativemismatch (- 1.5 ppm/fC). The highest compressive stressesfor the FC group (46.3 MPa) and the T discs (109.5 MPa)were associated with one of the negative mismatch cases(Aa = - 0.9). Shown in Fig. 4 are the mean values of biaxialflexure strength of slow-cooled, fast-cooled, and temperedspecimens for the three mismatch levels. Tempering treatmentof the compatible system (Aot = 0) resulted in the highestmean strength value of 116.2 MPa. Shown in Fig. 5 are themeasured biaxial flexural strength and the calculated compres-sive residual stress at the surface of body porcelain as a func-

tion of contraction mismatch for the tempered and fast-cooledspecimens. The surface stresses were calculated by use of thecrack-size data and an equation developed by Marshall andLawn (1978a).

Discussion.The initial crack length that develops at the surfaces of bi-

layered porcelain structures with variable levels of contractionmismatch can be reduced significantly by tempering. As shownin Table 2, the tempering treatment significantly reduced thecrack size compared with the slow-cooling and fast-coolingtreatments at both baseline and six months for all porcelaincombinations. Therefore, it can be inferred that tempering sig-nificantly increased the resistance to crack development andcrack propagation of bilayered porcelain structures, indepen-dent of the magnitude of contraction mismatch. The mean cracklengths of tempered specimens for the mismatch levels of - 0.9,+ 0.7, and + 3.2 ppmPC at six months were significantly greater(p <0.05) than those at baseline. Therefore, it appears thattempering stress was less effective in reducing the propagationrates of induced cracks.The crack lengths at six months (Table 2) indicate that the

influence of the cooling method was greater than that of ther-mal contraction mismatch. It appears that tempering stressestend to dominate incompatibility stresses at the surface of theporcelain, independent of the magnitude of contraction mis-match. Incompatibility stresses that develop because of con-

140 F

0-

z

cc

W

w

x

IL

4x

m

Fig. 4-Biaxial flexure strength of slow-cooled, fast-cooled, and tem-pered specimens.

X 120-

_10 0

zuW 80cc:

X 60wD 40w-JL. 2n

-2 .1 0 1 2 3 4 5

CONTRACTION MISMATCH (ppm/PC)Fig. 5-Measured flexure strengths and calculated surface stress due to

fast cooling and tempering.

a TSTRENGTH---_._--- T RESIDUAL

* FC STRENGTH----o--- FC RESIDUAL

IF

O. ' t-An"* ~~~~~~-

J Dent Res June 1991

MU

THERMAL TEMPERING OF FELDSPATHIC PORCELAIN

traction differences can have a significant overall effect onsurface stresses and the crack growth potential within thesesurfaces. However, the influence of cooling rate on the overallstress state within the surface was not a dominant factor in thedevelopment of compressive stresses until an extremely highcooling rate (tempering) was imposed.As shown in Fig. 4, the biaxial flexure strength for the

tempered porcelain as a function of contraction mismatch in-dicates that the body porcelain (control) group was associatedwith the largest strength value of 116.2 MPa. The biaxial flex-ure-strength values were calculated by assuming a Poisson'sratio of 0.28 (Farah and Craig, 1975) for dental porcelain.However, different values have been reported in other papers.KAse and Tesk (1985) utilized the sonic resonance techniqueto determine the elastic properties of two dental porcelains.They reported a Poisson's ratio of 0.19 for each porcelain atroom temperature. For determination of the influence of Pois-son's ratio on biaxial flexure strength, a parametric analysiswas conducted in this study. This analysis showed that a 50%reduction in Poisson's ratio for porcelain (with 0.14 used in-stead of 0.28) resulted in a maximum reduction of only 10%in surface-flexure stress values.The mean values of flexure strength for tempered and fast-

cooled specimens were higher for the negative mismatch group(- 1.5), compared with the positive mismatch group of + 3.2ppm/0C (Fig. 4). The data in Table 2 indicate that the cracklengths for the - 1.5 ppm/C mismatch groups were signifi-cantly smaller (p < 0.05) than those for the + 3.2 ppm/C mis-match groups in all three cooling conditions. Therefore, itappears that the results obtained from the microhardness in-dentation test correlate well with those from the biaxial flexure-strength measurements. This correlation is demonstrated in Fig.5. Flexure tests were conducted only for mismatch levels of- 1.5, 0, and + 3.2 ppm/0C. The mean values of measuredstrength (solid lines) and the calculated residual stresses (dottedlines) from Table 4 for tempered and fast-cooled specimens asa function of contraction mismatch were plotted in the samegraph. This similarity between the flexural-strength and com-pressive stress curves corresponding to each cooling conditionis an indication of the validity of the Marshall and Lawn equa-tion (1979). This equation can predict the amount of com-pressive surface stress induced by the tempering treatment andmay also be used as an indication of the strength enhancementeffect of tempering. Although the biaxial flexure-strength mea-surements were not made for the -0.9 and + 0.7 ppm/Cmismatch cases, these values can probably be approximatedfairly well from Fig. 5.The major conclusion of this study is that the resistance of

the porcelain surface to crack initiation and the overall strengthof the porcelain can be improved by tempering. However,relatively large variances of the data suggest that the temperingprocess is technique-sensitive. It should be emphasized thatthe analysis of surface stresses by use of microhardness inden-tation data was performed on 80 cracks induced in the surfaceof one specimen. Under this experimental condition, variationsin voids and porosities within specimens were not considered.Perhaps the experimental design could have been improved ifthe 80 indentations had been distributed across a number ofdiscs. Although feldspathic porcelain is generally used with ametal substrate, the opaque-body porcelain specimens in this

study were chosen to simplify the analysis of stresses producedby tempering. Further studies are continuing on metal-ceramicspecimens and tempering by ion exchange.

Acknowledgments.The authors gratefully acknowledge the assistance of Mr.

Paul Cascone of J.F. Jelenko & Co. for preparing the exper-imental porcelain formulations used in this study. The assis-tance of Ms. Anna Gray in the measurement of crack dimensionsis also appreciated.

REFERENCES

ANUSAVICE, K.J.; DEHOFF, P.H.; HOJJATIE, B.; and GRAY,A. (1989): Influence of Tempering and Contraction Mismatch onCrack Development in Ceramic Surfaces, J Dent Res 68:1182-1187.

ANUSAVICE, K.J.; GRAY, A.; and SHEN, C. (1991): Influenceof Initial Flaw Size on Crack Growth in Air-tempered Porcelain,J Dent Res 70:131-136.

BAN, S. and ANUSAVICE, K.J. (1990): Influence of Test Methodon Failure Stress of Brittle Dental Materials, J Dent Res 69:1791-1799.

CREYKE, W.E.C. (1968): Delayed Fracture of Glazed Porcelain,Trans Br Ceram Soc 67:339-365.

DEHOFF, P.H. and ANUSAVICE, K.J. (1989): Tempering Stressesin Feldspathic Porcelain, J Dent Res 68:134-138.

DOREMUS, R.H. (1973): Strengthening of Glass. In: Glass Science,R.H. Doremus, Ed., New York: John Wiley & Sons, Inc., pp.310-315.

DUNN, B.; LEVY, M.N.; and REISBICK, M.H. (1977): Improvingthe Fracture Resistance of Dental Ceramic, J Dent Res 56:1209-1213.

FARAH, J.W. and CRAIG, R.G. (1975): Distribution of Stresses inPorcelain-Fused-to-Metal and Porcelain, J Dent Res 54:255-261.

GARDON, R. (1980): Thermal Tempering of Glass, Vol. 5, Elasticityand Strength in Glasses. In: Glass Science and Technology, D.R.Uhlmann and N.J. Kreidl, Eds., New York: Academic Press, pp.144-216.

GUPTA, P.K. and JUBB, N.J. (1981): Post-Indentation Slow Growthof Radial Cracks in Glasses, JAm Ceram Soc 64:C112-C114.

JONES, D.W. (1983): The Strength and Strengthening Mechanismsof Dental Ceramics. In: Dental Ceramics, Proceedings of theFirst International Symposium on Ceramics, J.W. McLean, Ed.,Chicago: Quintessence Publishing Company, pp. 83-141.

KASE, H.R. and TESK, J.A. (1985): Elastic Constants of Two Den-tal Porcelains, J Mater Sci 20:524-531.

MARSHALL, D.B. (1980): An Improved Biaxial Flexure Test forCeramics, Ceram Bull 59:551-553.

MARSHALL, D.B. and LAWN, B.R. (1977): An Indentation Tech-nique for Measuring Stresses in Tempered Glass Surfaces, J AmCeram Soc 60:86-87.

MARSHALL, D.B. and LAWN, B.R. (1978a): Measurement of Non-uniform Distribution of Residual Stresses in Tempered Glass Discs,Glass Technology 19:57-58.

MARSHALL, D.B. and LAWN, B.R. (1978b): Strength Degradationof Thermally Tempered Glass Plates, JAm Ceram Soc 61:21-27.

MARSHALL, D.B. and LAWN, B.R. (1979): Residual Stress Effectsin Sharp Contact Cracking, Part 1. Indentation Fracture Mechan-ics, J Mater Sci 14:2001-2012.

WACHTMAN, J.B., Jr.; CAPPS, W.;, and MANDEL, J. (1972):Biaxial Flexure Test of Ceramic Substrate, J Mater 7:188-194.

Vol. 70 No. 6 1013