Dent Mater 8:203-207, May, 1992

The effect of glaze on porcelain strength C. W. Fairhurst, P. E. Lockwood, R. D. Ringle, W. O. Thompson 1

Dental Materials Section, Department of Restorative Dentistry, School of Dentistry, Medical College of Georgia, Augusta, GA, USA 1Office of Research Computing and Biostatistics, School of Graduate Studies, Medical College of Georgia, Augusta, GA, USA

Abstract. The self-glazing technique provides an esthetic and hygienic surface for crowns and fixed partial dentures that use porcelain veneers. A study of the biaxial flexure strengths of polished vs. glazed specimens is needed to verify that current laboratory methods are appropriate for planned fatigue studies. Four groups of 50 porcelain disk specimens each were subjected to the following polishing and firing procedures: group one was fired, glazed-no hold, and polished; group two was fired, polished, and glazed-no hold; group three was fired, polished and glazed1 min. hold; group four was fired, polished, and not glazed. The piston-on-three-ball method was used for testing biaxial flexure strengths. Significantly lower differences in biaxial flexure strengths were noted when group two values were compared with values from groups one, three and four. The results show that the Weibull distribution is an appropriate model for our studies. Differences in glaze thickness among the groups were noted in SEM examination; however, bulk (interior) microcrack density differences were absent. The specimens that were fired, polished to a 1 pm surface finish, and not glazed (group four) were significantly higher in flexure strength than groups one and three at the p < 0.001 level. The hypothesis that glazing of porcelain surfaces improves the biaxial flexure strength of test specimens was rejected.

INTRODUCTION Glazing of porcelain dental restorations is a routine procedure providing esthetic and hygienic glass-coated surfaces to the finished restoration. In this paper, only the self-glazing procedure is referred to and not the application of a lowfusing glass overcoat. Glazing is sometimes used to strengthen ceramics; however, the effectiveness of a glazing procedure on strengthening of porcelain restorations or appliances is uncertain. Binns (1977) has listed glazing under methods of strengthening glass. However, he also has questioned the effectiveness of this procedure because the surface of a dental porcelain appliance is often ground subsequent to glazing and still provides satisfactory clinical service. Glazing for the purpose of strengthening brittle glass composites can be thought of as the production of a surface layer of lower thermal expansion glass which upon cooling serves two functions: 1) it places the surface into a compressive state, and 2) the thin layer of glass reduces the depth and width of surface flaws and could theoretically strengthen the material. Corbitt et al. (1985) suggested that dental porcelain strength appeared to be controlled by intrinsic rather than surface flaws. In our ongoing research on porcelain fatigue, effects of

fatigue will be monitored by biaxial flexure testing of 1 mm thick porcelain disks. Surface preparation is of concern since specimens fail in tension when subjected to biaxial flexure testing. In the previous work of Morena et al. (1986), the "supporting tensile surfaces" were polished through 1 ~m diamond paste. This produces a flat test specimen resulting in surface flaws 1 ~m or smaller. Glazing is a routine dental laboratory procedure that further decreases the size of surface flaws. A comparison of the effects of polished vs. glazed surface on biaxial flexure strength seemed necessary prior to initiating the fatigue study. The purpose of this study was to test the hypothesis that the biaxial flexure strength of porcelain is significantly increased by glazing.

MATERIALSAND METHODS More than 200 gingival (body) porcelain disks measuring 12 mm in diameter and 1 mm in thickness (Jelenko Gingival - Lot #2012, Jelenko Dental Health Products, Armonk, NY, USA) were fabricated in a plastic die by use of vibration and blotting techniques. All specimens were fired in a programmable furnace (Sunfire 10, The J.M. Ney Company, Bloomfield, CT, USA) at a pressure of 0.010 MPa from 593°C to 927°C at 56°C/min. The pressure was returned to 1 atm at a temperature of 927°C, firing continued, and the specimens were removed from the furnace when the temperature reached 968°C. After this initial firing, all specimens were ground flat on both sides on a 70 ~m diamond abrasive steel backed wheel. Specimens for all groups were further ground (ECOMET-III, Buehler Ltd., Lake Bluff, IL, USA) on both sides through 15 ~m diamond paste. The side of the disk to be placed face down, bearing on the three stainless-steel ball bearings, will be described hereafter as the "supporting tensile surfaces" (STS) of the porcelain disks. The STS only were then further polished, in different sequences, through 1 pm diamond paste. Following the original firing and grinding sequence, some of the groups, as noted in the following paragraph, were given a glaze-firing which consists of an additional firing at 1 atm pressure at 56°C/min. from 593°C to 946°C, the temperature at which the specimens were removed from the furnace. Fifty specimens for each of the four experimental groups were randomly drawn from the common pool of once-fired specimens ground flat on both sides through 15 ~m diamond paste. Table 1 provides a summary of the surface treatments for each group. Group one specimens were glaze-fired, ground through 15 ~m diamond paste on both sides again, Dental Materials~May 1992 203

TABLE 1: SPECIMENGROUPSURFACECONDITION GROUP GLAZE-FIRED FINALTESTSURFACE 1 Yes lmm Diamond 2 Yes GlassyGlaze 3 Yes+lmin GlassyGlaze 4 No lmm Diamond

and had the STS polished through i pm diamond. Group two specimens had the STS polished through 1 ~m diamond, and then they were glaze-fired. Group three specimens had the STS polished through 1 ~m diamond, were then glaze-fired, and additionally held at the glazing temperature for 1 min. The presence of a glassy-appearing surface on the STS of groups two and three was visually confirmed. In previous work on the fatigue of porcelain, specimens were prepared with a single firing as specified by the manufacturer. A fourth group was prepared in order to compare the biaxial flexure strength data from groups one, two, and three to previously data from a study of single-fired specimens. This fourth group of specimens, as in the previous study (Morena, et al., 1986), was not subjected to a glaze-firing procedure. The STS of these specimens were polished through 1 ~m diamond after the original firing and grinding sequence. The thickness and diameter of all disks were measured with a digital micrometer (Mitutoyo Digimatic Model No. 293, MTI Corporation, Paramus, NJ, USA) equipped with 3 mm diameter spline attachments that reduce the disk area measured. The thickness was measured at five places with the center of the disk within the area of the measuring anvils. The thickness of a single specimen was nominally 1 mm with a standard deviation of 0.02 mm. The mean thickness for the 50 samples from each group one through four respectively was 1.01, 1.05, 1.06 and 1.01 mm. The diameter of the disks was measured at two places and the nominal mean was 12.0 mm and a typical mean deviation of 0.07. The combined variability in thickness and diameter was calculated to account for a possible error of approximately one percent of the biaxial flexure strength value. All specimens were stored until tested in closed plastic containers under ambient conditions at relative humidities between 30 and 50 percent. All specimens were tested in circulating 37°C water and were loaded at a constant stressing rate of 1.7 MPa/s in an Instron 8562 (Instron Corporation, Canton, MA, USA) servomechanical testing instrument. The test fixture has been described in detail elsewhere (Radford and Lange 1978) and conforms to ASTM Standard Designation: F394-78, 1978 with two exceptions: 1) the piston end was modified by cutting a concave hemispherical recess in the end of the piston rod and inserting a stainless steel ball with a flat surface (Radford and Lange, 1978) that conforms to the specimen surface and avoids uneven loading, and 2) as a result of this piston modification, no pad of non-rigid material was used between the loading surface and the disk. The diameter of the loading piston's flat surface in contact with the disk were 0.780 +0.005 mm. The 1.59 mm ball bearing supports are stainless steel. The radius of the support circle formed by the three stainless steel balls was 3.16 +0.01 mm. Only one fixture was used in this study, and systematic equipment errors were estimated at 0.75% of the biaxial flexure strength value. The merits of the piston-on-three-ball biaxial flexure test have been studied and discussed by several investigators (Shetty et al., 1983; Ritter et al., 1980; Radford and Lange,

204 Fairhurst et aL/Effect of glaze on porcelain strength

1978; andWachtmanetal., 1972). The primary virtue ofthe test fixture used in this study is that the lack of perfectly plane and parallel surfaces has been compensated for in the fixture design. All specimens were examined after fracture to determine the type of break. Nearly all specimens broke into three or more pie-shaped pieces with the specimen center as the common fracture junction. Those seven specimens that did not break in this manner, broke into two unequal pieces, one large pie-shaped piece with the tip at the center under the piston and the matching second piece. This was taken to indicate that the fracture initiation site was under the piston where the stress was at a maximum. As described previously in this report, the piston contact surface was swiveled to assure uniform contact under loading; thus, uneven edge contact on the compressive surface of the disks was unlikely. The observed fracture loads and specimen dimensions were used in the calculation of stress at fracture as described in ASTM Standard Designation: F394-78 for biaxial flexure strength. A value for Poisson's ratio of 0.25 was used in the calculations. The fracture stress distributions were hypothesized to be Weibull distributions and were tested for the degree of fit to the theoretical Weibull distribution using the KolmogorovSmirnov one-sample test. The Kolmogorov-Smirnov twosample test was used to assess distribution differences between sample groups. A scanning electron microscope (SEM) was used to obtain photomicrographs of random broken test specimens from each of the four groups. The glazed surface of each test specimen was etched for 30 s with 1.25 percent hydrofluoric acid (HF).

RESULTS A Weibull cumulative distribution histogram for group 3 is shown in Fig. 1, where the probability of fracture is plotted vs. the stress at fracture for each sample in group three. The theoretical Weibull distribution given by the constants of the two-parameter Weibull distribution is shown overlaid as the solid curve. Fig. 2 represents the same type of data for group two and shows a smaller modulus and scale factor than the distribution for group three shown in Fig. 1. The theoretical distributions calculated from Weibull constants of the four groups tested are shown plotted together in Fig. 3. Statistics describing biaxial flexure strengths are shown in Table 2. The modulus and scale factors are the Weibull shape and scale factor, respectively. The Kolmogorov-Smirnov (K-S) one-sample test provides a goodness-of-fit test of the data to the Weibull function. The p-values are listed where smaller p-values indicate a poorer fit of the data to the most appropriate Weibull model. The next two values are for biaxial flexure stress-at-fracture (s) values at 0.50 or the median probability of failure, and below that, the average strength (s). SD represents the standard deviation. The value of the coefficient of determination, r 2,is from the linearlog form of the Weibull distribution function, where LnLn (1/(I-P)) vs. Ln(biaxial strength) follows a simple linear relationship when the Weibull model is appropriate. Each group contained 50 specimens. The hypothesis that two Weibull strength distributions were equivalent was tested using the Kolmogorov-Smirnov two-sample test. Groups one and three were not different (p = 0.997); however, groups one and three were significantly

WEIBULL STRENGTH DISTRIBUTION OF JELENKO BODY PORCELAIN i

I

r

TABLE 2: BIAXIAL STRENGTH OF JELENKO PORCELAIN TREATMENT GROUPS Strength 1 2 3 4 Parameters (F-GO-P)* (F-GO-P)* (F-P-G1)* (F-P-NG)*

,

n

-G

0.8 -S

.= mO

0.6

.~ 0.4 L 0.2

-

Modulus"

14.7

11.0

14.3

14.3

Scale Factor= K-S (p-value)

49.3 0.09

47.2 0.68

49.0 0.99

53.4 0.98

s(MPa) (.5) s(MPa) (Mean) SD

46.48 47.65 3.77

44.76 45.27 4.26

47.54 47.24 4.04

52.30 51,56 4,10

r2 0.98 0.95 0.99 0.98 N 50 50 5O 50 "Wiebull shape factor, ~Wiebull shape factor, K-S = Kolomogorov-Smirnov, s = Stength.*Groups were treated in the order labeled above as follows: F- fired GO- glazed (no hold); NG- no glaze; P- polished;G1- glazed (1 min hold)

I

,

,

,

,

35

L

,

,

,

,

40

;

I

I

45

50

55

Stress

at

Fracture

,

,

60

(MPe)

Fig. I. Acumulativedistributionhistogram,wheretheprobabilityoffracturevs.the stressat fracture for each sampleis shownfor groupthree. The curveis the bestfit Weibull distributionderived from the modulus and scale factor constantsof the Weibulldistributionof the samples. WEIBULL STRENGTH DISTRIBUTION OF JELENKO BODY PORCELAIN -G M

0.8

-S

®

O6

o ~

0,1

a.

0,2

,

,

,

J

I

39

43 Stress

,

,

,

I

,

,

L

47 at

Fracture

,

,

J

I

51

55

(MPa)

Fig.2. TheWeibulldistributionofgrouptwo.Thisdistributionhadasmallermodulus than groupthree. Weibull Strength Distributions of J e l e n k o Body P o r c e l a i n

GROUP GROUP GROUP GROUP

o8

1 2 3 4

/ /

,'' / / /

/

/

,

~

/

/

/

02

/"

2//''" / //!'/'//

06

04

/// /"

/

different from group two (p < 0.02). The group four distribution was significantly different and higher in strength values than the other three groups, p < 0.001, Fig. 3. In order to better understand the effects of the two glazing procedures on specimen test surfaces, SEM examination of fractured disk sections was carried out. Fig. 4 is a photomicrograph of the etched surface of a specimen from group two which shows a typical leucite structure. This leucite structure was not evident on glazed samples prior to etching. Fig. 5 is a photomicrograph of the 30-second-etched surface of a specimen from group three, the group held at the glaze temperature for 1 min. The etch appears to have removed enough of the glass to just reveal the leucite. The specimen shown in Fig. 5 was given an additional 30 s HF etch for a total etching time of 60 s. Fig. 6 shows the resulting surface that reveals the leucite structures in greater detail. A longer etching time for the group three specimen was required in order to remove the glassy glaze layer of glass on the surface. This indicated that a thicker coat of glass must have been formed during the 1 min hold at glaze temperature. Fig. 7 shows a section from within the body of one of the specimens from group one. No clear quantitative differences in the microcrack density of specimens was evident when photomicrographs from each of the four groups were compared. Microcracks occurred in this system due to differences in thermal coefficient of expansion between the leucite crystals and matrix glass. It should be noted that in most of the sections examined, microcracks occurred around clusters of small leucite crystals and only occurred around single leucite particles when these particles were exceptionally large. The microcracking around large particles in this porcelain system was not unexpected (Davidge and Green, 1968). However, it appears that agglomerates of small particles can act in a similar manner to a single larger particle. Both of these phenomena can be observed in Figs. 6 and 7. These observations lead one to question whether individual leucite particle size is the controlling factor in the formation of intrinsic microcrack flaws since the observation is that clusters of very small particles also appear to promote microcracking.

/"

DISCUSSION 0 t

35

,

,

,

,

I

40

,

,

,

,

L

45

,

,

,

,

i

,

,

,

50

Stress at Fracture (MPa) F/g. 3. CumulativeWeibulldistributionsof all four groups.

,

I

55

,

,

,

,

I

60

The data suggest that a single firing of body porcelain disks adequately represents a dental body porcelain that is used for porcelain-fused-to-metal restorations. The mean strength of once-fired specimens is eight percent greater than the Dental Materials~May 1992 205

Fig.4. Photomicrograph of the surface of a specimen from grouptwo showing typical leucite structure after a 30 s etch.

Fig.6. PhotomicrographofthespecimeninFigure5afteradditionaletchof3Os. The leucite structure is completely revealed indicating that the glaze layer was thicker than that of the group two specimen.

Fig.5. Photomicrographofegroupthreespecimen. The3Osetchhasrevealedsmall portions of the leucite. This specimen had a thicker glaze layer than the specimen in Figure 4,

Fig. 7. Photomicrograph of a section within the body of the specimen from group one. Examination of sections from the other three groups did not demonstrate any qualitative difference in microcrack densities among the four groups.

mean strength of twice-fired specimens, and this difference is significant at the p < 0.001 level. The group two distribution does not fit the Weibull model (p = 0.68) as well as groups one, three, and four, respectively, with p values of 0.90, 0.99, 0.98. The biaxial strength distribution of the porcelain is a physical measure of flaw size distribution in the specimens. Group two has a different distribution and, therefore, a different flaw size distribution than the other three groups. Although the mechanism of formation of the larger flaw size during the glaze firing is not understood, the glaze treatment introduces large flaws into specimens, resulting in lower strengths than in the other groups. Whereas, the glaze firing plus a one minute hold at the glazing temperature may have provided time for the larger flaws to heal. Groups one and three had identical strengths and distribution parameters and similar thermal treatments, but

different surface finishes. Group three was held at glaze temperature for one minute, then tested; group one was glazed with no hold and polished before testing. Thus, it is not evident that surface damage-induced stress is solely responsible for higher strength than undamaged surfaces (Cook et al. 1981; Giordano et al., 1991) or that reduction of surface damage via the glaze-firing process is the factor responsible for strength reduction in these feldspathic ceramic specimens. As can be seen by comparing groups one and four, the strength of group four is higher than group one, yet the surface finishes are identical, and the only difference is their thermal treatment. Regardless of the order of glaze and other surface treatments, the second-fire glaze treatments of groups one, two, and three produced strength values significantly lower than the strength of the single-fire group four. One cannot help but wonder why this occurred. Obviously, the possible

206 Fairhurst et aL/Effect of glaze on porcelain strength

advantages from the second-fire with respect to flaw-healing, annealing out polishing stresses, etc., are not realized in comparison to group four, since the second firing weakened the material. The effects ofmultiple firings on leucite volume fraction and microcrack density have been studied by Fairhurst et al. (1980); Mackert and Evans (1991); Mackert et al. (1991a; 1991b). This thermal treatment effect may be partially responsible for the strength differences observed in this study as well as in other studies and cannot be dismissed as the effect of a simple surface anneal. It is interesting to note that Rosenstiel et al. (1989) found that polishing, when compared to glazing, produced a surface with a significantly higher fracture toughness. In the beginning of this discussion, the reader's attention was directed to the strength differences between the once-fired group four and the other groups. These differences are apparent in Figure 3 which shows the cumulative Weibull distribution curves for all four groups. The porcelain used in this study is a good representative of the Weinstein type feldspar porcelains reported on by Twiggs et al. (1990). In testing such porcelains, the flexural strength can be altered by glazing techniques; however, when specimen surfaces are finished by polishing as in this study, fracture is controlled by the intrinsic flaw distribution. This intrinsic flaw distribution is a result ofthe locations and nature ofthe leucite particles surrounded by a lower thermal expansion glass matrix which forms microcracks upon cooling, due to differences in thermal expansion coefficients. This process is well described in the literature (Davidge and Green, 1968). However, it is not clear whether with increasing microcrack density there is an accompanyingincrease in strength due to "crack stopping" or a decrease in strength due to the higher probability of producing a large critical flaw, as may have occurred in group two. It is also possible that certain thermal effects may blunt microcracks. In summary, the results of this analysis suggest that in addition to details of specimen surface finish, the effectofthe number offirings, duration offirings, and the firing temperature must all be explicitly provided in order to compare properties offeldspathic porcelain with other existing physical property data. The self-glazing procedure is appropriate for clinical use since it provides smooth hygienic surfaces as clearly shown in the photomicrographs of the test surfaces. However, further study is needed to determine the effect of the duration of glaze hold time on fracture properties. The hypothesis that glazing porcelain surfaces strengthens the test specimen for biaxial strength is rejected. Some glazing techniques can be detrimental to the fracture properties of leucite-containing porcelains. The number, duration, and temperature of specimen firings in part control the fracture strength of porcelain ceramics.

Address correspondence and reprint requests to: C. W. Fairhurst Medical College of Georgia, Dental Materials Section Department of Restorative Dentistry, School of Dentistry Augusta, GA 30912-1264 USA

REFERENCES

ASTM Designation: F394-78 (1978). American Standard Test Method for Biaxial Flexure Strength (Modulus of Rupture) of Ceramic Substrates, in Annual Book of AST Standards. Philadelphia, PA, American Society for Testing and Materials (reapproved 1984), 434-440. Binns DB (1977). The Physical and Chemical Properties of Dental Porcelain, In: Dental Porcelain: The State of the Art- 1977, 25-34. Cook RF, Lawn BR, Dabbs TB, Chantikul P (1981). Effect of Machining Damage on the Strength of a Glass-Ceramic, J Am Ceram Soc / Communication 64(9):C-121-C- 122. Corbitt GV, Morena R, Fairhurst CW (1985). Fracture Stress of a Commercial Dental Porcelain and Its Components, J Dent Res 64:296 Abst. No.1089. Davidge RW, Green TJ (1968). The Strength of Two-Phase Ceramic/Glass Materials, J Mater Sci 3:629-634. Fairhurst CW, Anusavice KJ, Hashinger DT, Ringle RD, Twiggs SW (1980). Thermal Expansion of Dental Alloys and Porcelains, J Biomed Mater Res 14:435-446. Giordano R, Cima M, Pober D (1991). Effects of Surface Finish on Strength of Various Dental Ceramics, J Dent Res 70:433 Abst. No.1340. Mackert Jr JR, Evans AL (1991). Effect of Cooling Rate on Leucite Volume Fraction in Dental Porcelains, JDent Res 70(2):137-139. Mackert Jr JR, Evans AL, Twiggs SW (1991a). High-Ternperature X-Ray Diffraction Measurement of Sanidine Thermal Expansion, J Dent Res (Special Issue) 70:456 Abst. No.1520. Mackert Jr JR, Rueggeberg FA, Lockwood PE (1991b). Isothermal Anneal Effect on Microcrack Density Around Leucite Particles in Dental Porcelains, JDent Res (Special Issue) 70:456 Abst. No.1519. Morena R, Beaudreau GM, Lockwood PE, Evans AL, Fairhurst CW (1986). Fatigue of Dental Ceramics in a Simulated Oral Environment, J Dent Res 65(7):993-997. Radford KC, Lange FF (1978). Loading (L) Factors for the Biaxial Flexure Test, J A m Ceram Soc 61(5-6):211-213. Ritter Jr, JE, Jakus K, Batakis A, BandyopadhyayN (1980). Appraisal of Biaxial Strength Testing, JNonoCryst Sol 38 & 39:419-424. Rosenstiel SF, Baiker MA, Johnston WM (1989). A Comparison of Glazed and Polished Dental Porcelain, Int J Prosthodont 2(6):524-529. Shetty DK, Rosenfield AR, Duckworth WH, Held PR (1983). A Biaxial-Flexure Test for Evaluating Ceramic Strengths, J A m Ceram Soc 66(1):36-42. ACKNOWLEDGMENT Twiggs SW, Hashinger DT, and Fairhurst CW (1990). VisWe appreciate the review of this manuscript by Dr. cosities of Porcelains Formulated from the Weinstein Robert Morena, Corning Inc., Corning, NY, USA and Delora Patent, J A m Ceram Soc 73(2):446-449. Hashinger, N. Augusta, SC, USA. Wachtman JB Jr, Capps W, and Mandel J (1972). Biaxial Flexure Tests of Ceramic Substrates. J Mater 7(2):188Received November 26, 1991/Accepted January 29, 1992 194.

Dental Materials~May 1992 207

The effect of glaze on porcelain strength.

The self-glazing technique provides an esthetic and hygienic surface for crowns and fixed partial dentures that use porcelain veneers. A study of the ...
881KB Sizes 0 Downloads 0 Views