Dent Mater 8:149-152, May, 1992
Strengthening of porcelain by ion exchange subsequent to thermal tempering K.J. Anusavice, C. Shen, B. Vermost, B. Chow
Department of Dental Biomaterials, College of Dentistry, University of Florida, GainesviUe, FL, USA
Abstract. The purpose of this study was to test the hypothesis that high-leucite content feldspathic porcelain can be strengthened best by thermal tempering followed by ion exchange, compared with either thermal tempering or ion exchange alone. The results of biaxial flexure testing indicate that thermal tempering was the most effective way of strengthening the porcelain among the three methods tested. The combined procedure did not yield the greatest strengthening effect as hypothesized, but one that is only slightly stronger than that produced by ion exchange alone. Weibull analysis indicated that the ion-exchange treatment resulted in a narrower distribution of biaxial flexural strength compared with that of untreated or thermally tempered porcelain. The narrower distribution of biaxial flexural strength is considered desirable for enhancing the longevity of ceramic-based dental restorations.
INTRODUCTION It is well established that the presence of surface flaws reduces the effective strength ofglasses and ceramics. Many attempts have been made to achieve tensile strengths closer to the theoretical value calculated from intrinsic cohesive forces by producing flaw-free glass. Etching of glass and ceramic surfaces reduces the influence of surface flaws, which leads to an increase of the strength in the glass, but is not deemed practical ifa chemically aggressive environment is required. Another approach is to produce a more desirable stress distribution within the surface by a thermal anneal treatment. By far, the most effective method of strengthening the glass is to generate a state of compression in the surface. This can be accomplished either by physical tempering or chemical strengthening. Physical tempering is a process whereby a glass article is heated to a temperature below its softening temperature and then quenched rapidly to room temperature, usually in air, molten salt or oil. During the rapid quenching, the exterior of the glass quickly becomes rigid while the interior cools more slowly and continues to contract after the exterior has become hard. Subsequent interior contraction relative to the surface produces compression in the outer surface while the interior is in a state of tension. The presence of this surface compression has been shown to improve the biaxial flexural strength of a body porcelain by as much as 158% (Anusavice and Hojjatie, 1991). Chemical strengthening is an ion exchange process that alters the chemical composition by introducing larger alkali metal cations in the surface layer of the glass while the bulk
of the glass retains the original composition. After cooling, compressive stress develops in the surface region that has changed chemically. Two different mechanisms have been suggested to explain the development of the compressive stress. Hood and Stookey (1957) exchanged lithium ions for sodium ions in the surface layer, resulting in glass with a surface oflower coefficient ofthermal contraction. A favorable stress profile was thus produced upon cooling. Kistler (1962) replaced sodium ions in the surface with larger potassium ions below the transformation range of the glass; the viscosity was so high in that temperature range that stress induced due to crowding of potassium ions in the places of smaller sodium ions could not be relieved. This also yielded a layer of compressive stress in the surface. A previous study (Anusavice et al., 1990) of the effect of thermal tempering and ion exchange on seven dental porcelains showed that both methods are effective in enhancing the strength of dental porcelains. Thermal tempering produced a deeper layer of compressive stress which yielded a greater increase in flexural strength compared with that associated with the ion exchange process. Ion exchange produces a state of compression to a limited depth from the surface, ranging from 30 to 100 pm, and a smaller degree of improvement in biaxial strength is realized. However, the magnitude of the compressive stresses in the surface region of ion-exchanged specimens was much greater than that ofth ermally tempered specimens. In the development of dental porcelains, various attempts have also been made to improve the bulk strength of the material. Those include the use of a higher expansion metal substrate (metal-ceramic), dispersion strengthening (aluminous core material), and controlled crystallization (Dicor castable ceramic, Dentsply, York, PA, USA). It has been shown that crystalline leucites (K~O"A1203"4SiO2)can precipitate from a feldspathic glass matrix with high potassium content, and function as a reinforcing agent of the porcelain. This approach was put to practice by Katz (1989) and a commercial product named Optec HSP (Generic/Pentron, Wallingford, CT, USA) has since become available and has been shown to exhibit a higher strength compared with other commercial porcelains (Katz, 1989). To encourage precipitation ofleucite from feldspathic glass matrix, the potassium-to-sodium ratio is increased. This composition change raises a question as to how effectively leucite-containing porcelain with a reduced sodium content can be chemically strengthened, or physically strengthened
Dental Materials~May 1992 149
in the presence of a reduced volume fraction of glass matrix and an increased concentration ofleucite crystal. It is possible that a combined procedure of ion exchange and thermal tempering can more effectively strengthen porcelain than either method applied alone. The purpose of this study was to test the hypothesis that, despite the lower sodium content, either tempering or ion exchange can strengthen the high-leucite content porcelain, and that a combined treatment ofthermal tempering followed by ion exchange can strengthen the porcelain to generate a greater level than that achieved by either method alone.
MATERIALS AND METHODS Forty-two disks ofOptec HSP porcelain specimens (shade 3.5), 16 mm in diameter and 2.5 mm thick, were prepared. A cylindrical brass mold, approximately 3.5 mm deep was used to form the porcelain disks specimens prior to firing. After the first firing, the specimens were ground to a thickness of 2 mm starting with 60 grit and finishing with 600 grit abrasive paper and then polished through 1 ~m A1203 abrasive prior to glazing. Each disk was glazed at 1016°C and slowly cooled (SC) to room temperature by terminating power to the furnace. Six disks were used as controls. The remaining disks were divided into six groups of six disks each and subject to one of the following six treatments: (1) thermal tempering by forced convective cooling in air (T) at a pressure of 0.34 MPa after a heat soaking at 850°C for 2 rain, (2) a tempering control treatment (TC) consisting of the same heat soaking treatment as in (1) followed by slow cooling in the furnace, (3) ion exchange (I) at 450°C in the furnace for 30 rain using a slurry of the ion-exchange agent Tuf-Coat (GC International, Tokyo, Japan) applied on one side and then fast cooled in ambient air, (4) an ion exchange control treatment consisting of the same heating and cooling cycle as the ion exchange group without application ofion-exchange agent (IC), (5) a combined tempering and ion exchange treatment (TI) consisting of the same heat soaking and rapid cooling cycle used for the tempered group and followed by the same treatment as the ion-exchange specimens, and (6) a tempering-ion exchange control treatment (TIC) produced by combined tempering and ionexchange procedures but without the application of ionexchange agent. Ion exchange was carried out by applying a thin layer of the ion-exchange slurry to the polished surface of the specimen and then skimmed to a uniform thickness with a knife blade while the specimen was stabilized in the trimming fixture at a fixed depth of approximately 250 lam below the fLxture surface. Each of the slurry-coated discs was then placed on a tray to dry for 20 min at 150°C in a small oven. After drying, specimens were heat-treated at 450°C for 30 min in a Unitek burnout furnace and then fast cooled in air. The residue was brushed and rinsed off under running water. All seven groups of specimens were stored in deionized water at 37°C for 7 d prior to testing and analysis. Biaxial flexural strength was determined for six discs per group using a pin-on-three-ball f~xture (Wachtman et al., 1972). The specimens were loaded to failure under water (37°C) at a crosshead speed of 0.5 mm/min using a universal testing machine (Instron Model 1125, Instron Corp. Canton, MA, USA). The load was centrally applied using a flat-ended indenter, with the treated or glazed surface of the disk seated on the three-ball support. The biaxial flexural strength data calculated by an equation proposed by Kirstein and Woolley
150 Anusavice et aL/Strengthening of porcelain
(1967) were also fitted by nonlinear regression into the twoparameter Weibull distribution function, which is expressed as follows: Pf = 1 - exp - G/G o (1)
[(
)m]
where Pf is the probability of porcelain failure under a stress, G is the biaxial flexural strength calculated from the failure load, G0is the characteristic strength which is for normalizing the stress term and has no physical meaning, and m is the shape factor. The calculated biaxial flexural strength values of the test group were first sorted in ascending order and a value of probability of failure (< 1.0) was then assigned to each strength in the order of 1/(n+ 1) to n/(n+l), where n is the total number of specimens. Since there were six specimens for each group, the probability of failure was 14.3% for the specimens with lowest strength and 85.7% for those with highest strength. The values of Goand m ofeach group were determined by fitting the biaxial flexural strength and the assigned probability of failure into Eq. (1) by a nonlinear regression procedure. RESULTS The mean biaxial flexural strength values and their 95% confidenceintervals for the seven treatment groups are shown in Table 1. The improvement of strength ranges from a negative 5.2% for the ion exchange control to a positive 43.5% for tempered specimens. Analysis of variance indicated that a significant difference (p < 0.01) existed among groups. Results of Tukey's HSD test indicated that there were no statistically significant differences (p > 0.05) among the four control groups, between groups T and TI, and between TI and I. The results of Weibull analysis are shown in Table 2. The value of m, which is shown with its respective 95% confidence interval, indicates the distribution ofbiaxial flexural strength TABLE 1: BIAXlAL FLEXURALSTRENGTHAS A FUNCTIONOF TREATMENT Treatment
Code
Strength(aPa)
Slow Cooled Tempered Control Tempered Ion Exchanged Control Ion Exchanged
SC TC T IC I
143.2 + 142.5 + 205.5 + 135.8 + 168.9 +
Tempered & Ion
TIC
152.8 + 21.8
13.8 11.7 15.5 7.3 16.9
% Change vs. SO 0.0 -0.5 I
I
43.5 -5.2 18.0 6.7
Exchanged Control Tempered & Ion Exchanged TI 183.2 + 14.6 IJ 27.9 * Vertical lines connect mean values that are not statistically different at p = 0.05.
TABLE 2: RESULTS OF WEIBULL ANALYSIS AS A FUNCTIONOF TREATMENT Characteristic Treatment Code Strength (% Mea) Slow Cooled SC 150.1 Tempered Control TC 147.7 Tempered T 218.5 Ion Exchanged Control C 139.4 Ion Exchanged I 176.9 Tempered & Ion Exchanged TIC 162.9 Control Tempered & Ion Exchanged TI 186.4
Shape Factor
Goodnessof-Fit
(m)
( °/o )
10.8 _+6.7 10.7 + 3.1 8.7 + 6.4 17.4 +11.0 14.3 + 7.4 7.2 + 2.2
91.1 97.4 89.8 90.9 94.5 97.6
14.6 + 9.9
91.4
100.9:~
,' // 4 :3/d
Slow Cooled
:'
0.8
• TemperedContro~
0.7
:'
i
Ion Exchanged Control
"6
~
/~-~
,': /~/
Ion ExchangedContro,
0.6-
I/ //
!
Tempered& _.
values obtained experimentally. Fig. 1 shows the experimental data along with best-fit Weibull distribution for the four controls. Fig. 2 shows the experimental data and Weibull distribution of three treated specimens compared with that for the slow-cooled group. Fig. 3 shows the experimental data and Weibull distribution of steps from slow cooled to the combined procedure. The experimental data showed that the ion exchange process resulted in a narrower distribution of strength values while the tempering process produced a broader distribution of strength values.
: /i
/
/
/
! Oli~%
0.5-
id
; ~E/
o
0.3
th
@
/// Z':"
"~ 0.4.0
0.2
DISCUSSION
0.12Y,:" 0.0-
i
r
i
i
r
50
75
100
125
150
E 175
Biaxial Flexural Strength,
i
i
200
225
MPa
Fig. 1. Effect of cooling cycle on the flexural strength of porcelain with a high-leucite content. 10
/I
09.... 0.8
Slow Cooled • Tempered
0.7
Ion Exchanged
--
Tempered & Ion Exchanged
0.6-
"6
t
---
0.4
:}'
t
/
/
f
/ [ .I
01-
/"
, ." •
/
/
/ I
/
'
,*o
it /
/
02-
o," ,: .'
I
'
.0 0 13- 0,3~
:
I
i,/I
"'"'"
• ,, ,"'"
I
J / I /'
05
/' "
iI
"/"--'--'T / /
, / /'
/
./
iI / /
," "//,/:
/
/ /" //.-"
00-
75
5O
,00
~25
~so
,Ts
2oo
Biaxial Flexural Strength,
22s
250
275
MPa
Fig. 2. Effect of treatment on the flexural strength of porcelain. 10
/
09 - I) 0.8 (1)
B
~
0.7-
-
,! / I / /' / ~1 //~
-- o Tempered & Ion
--
[
Ion Exchanged
.z~
0.4-
/
,,> i/
4/ /
~
"..?
.
,,"" ,'"
:"
/
//
,"
,,""
/~
,' // .
./:"/:
/...":
i? "
0.2-
~./
I
J/ .(/~
~-- 0.3-
:" +,' ,, e."
/
!~/
"6 0.5-
"
/
i iI"
:) Tempered &
..'"-- - -
II / ?/ I '~
[I!
Slow Cox)led
- • Tempered Exchanged Control
0.6-
~" ---/-~7~
,I
//
0.0~
75
,oo
,2s
,50
,75
~0
Biaxial Flexural Strength,
22s
~50
27s
MPa
Fig. 3. Biaxial flexural strength of porcelain as influenced by the sequence of
According to a published patent (Katz, 1989), the leucite content of the Optec porcelain is at least 45 w/o and typically can be within the range of 55 to 75 w/o. Assuming that the composition disclosed in the patent description is the same as the material used in the study, i.e., a porcelain that contains 45 w/o ofleucite, for 100 g ofOptec HSP, the approximate mole ratio of SiO2 : A1203 : K20 : Na20 : Li203 : CaO in glassy phase of the porcelain is 0.57 : 0.06 : 0.05 : 0.06 : 0.05 : 0.02. This analysis shows that there are more lithium and sodium than potassium in the glassy matrix. This chemical makeup may favor a situation in which Na ÷replaces Li÷, while K÷replaces Na ÷duringion exchange treatment. The 18% improvement in the biaxial flexural strength value is low compared with the range of 31.9% to 96.8% reported previously for seven commercial porcelains of higher sodium content (Anusavice et aI., 1990). It has been suggested that the existing K÷within the structure can reduce the driving force for potassium uptake near the surface region and result in a lower percentage of improvement of the failure strength values (Piddock et al., 1991). It is important, nonetheless, to note that even a high leucite containing porcelain can be strengthened by ion exchange as long as smaller monovalent ions are present. The four control groups were heat soaked and then cooled from the following temperatures: 1050 ° C for slow cooling, 850°C for the tempered control, and 450°C for the ion exchange and combined treatment control. It is reasonable to assume that an equilibrium state was being established prior to cooling and the temperature of heat soaking influenced the stress distribution in the porcelain surface upon cooling. Weibull analysis indicated that the group cooled from the lower temperature had a narrower distribution of flexural strength, despite the fact that the mean biaxial flexural strength of the groups was not statistically different (p > 0.05). The ion exchange control group, which was fast-cooled to room temperature from 450°C, exhibited the highest m value. This implies that 450°C is low enough for Optec HSP to avoid development of a nonuniform stress in the surface which is often associated with rapid cooling from a higher temperature. Fig. 2 shows the effect of treatments on the distribution of flexural strength of the porcelain. The tempered group has the highest mean strength value and the widest distribution of the four groups. It indicates that, although tempering is the most effective method of strengthening, the results are less predictable because of a wider distribution. Strengthening by ion exchange is less effective but the distribution of failure strength values is much narrower. This may not appear to be of significance if one compares only the average strength values among groups. In reality, if we assume that a failure rate of I to 5% is considered acceptable, the average strength values shown in Table 1 become meaningless. From the point of designing a prosthesis, the failure strength corresponding to
combined treatment.
Dental Materials~May 1992 151
failure level of 5% or less is more meaningful. Table 3 shows the predicted strength at 1% and 5% of failure by Weibull distribution for all seven groups. At 5% of failure level, the strength values are 155,144 and 152 MPa for tempering, ion exchange, and tempering and ion exchange, respectively. At the 1% failure level, the strength values are 128, 128 and 130 MPa for tempering, ion exchange, and tempering and ion exchange, respectively. It becomes clear that a higher mean value of strength, such as that produced by tempering, is not necessarily equivalent to a higher resistance to fracture when the acceptable failure level is set at 1% or 5%. Fig. 3 shows the changes in the biaxial flexural strength value at each step from slow cooling to tempering and ion exchange. The pattern of changes implies that ion exchange treatment at 450°C for 30 rain was not sufficient to produce an equilibrium stress state, but to cause significant stress relaxation within the tempered layer. Meanwhile, the application of ion-exchange paste to the stress-relaxed disk surfaces yielded a strength value intermediate to those associated with ion exchange and tempering alone. It is interesting to note that the shape factor of the tempered, and tempered and ion exchange control specimens is about the same, while ion exchange improves the shape factor (Table 2). This indicates that the tempering procedure yields a wide distribution of strength values and the ion exchange control cycle can only reduce the magnitude but not the variability of the surface stress within the tempered layer. We may conclude from Figs. 1 and 3 that 450°C is low enough to avoid development of nonuniform stress from coolingbut not high enough to relieve tempered stress completely. The number of specimens used by investigators for a twoparameter analysis of the Weibull distribution is variable. McCabe and Walls (1986) argued that the value ofm gives an indication of the reliability of the flexural strength; therefore, the 95% confidence interval of the m value calculated from experimental data should be as small as possible. They concluded from their study that a minimum of 30 would be needed to perform a meaningful analysis by two-parameter Weibull distribution. In fact, the m values shown in Table 2 are not statistically different for all seven groups by Tukey's HSD because of the relatively wide range of 95% confidence interval of the m value of each group. Since the strength values of the four control groups of the present study were not significantly different (p > 0.05), we can treat the strength values offour control groups as one group of 24 specimens, and fit the value along with respectively assigned probability of failure based on 24 specimens into equation (1). The estimated values of Go and m are 148 MPa and 11.1 _+ 1.2, respectively. The m value calculated from a population of 24 specimens has a smaller 95% confidence interval compared with that of the individual control groups. This result is consistent with the conclusionofMcCabe and Walls (1986). In the case of lifetime predictions, there are more variables involved and a minimum of 100 or more has been suggested by Ritter et al. (1981). Obviously, if the m value obtained is for the purpose of predicting the reliability of a particular material, a higher number of specimens is needed to assure a small range of 95% confidence interval. The m values calculated in this study, which are based on 6 specimens with a goodnessof-fit of 90% or greater, should only be used for the purpose of comparing the effectiveness of treatments on each particular porcelain. Further studies should concentrate on the analysis of the
152 A nusavice et aL/Strengthening of porcelain
temperature and time for which degree of stress relaxation will be least affected while significant strengthening by ion exchange can still be achieved. It is also conceivable that additional strengthening can occur by changing the sequence of treatment to ion exchange followed by tempering. DeHoff and Anusavice (1989) showed by finite element analysis that rapid cooling from 650°C can be effective in strengthening feldspathic porcelain. Since ion exchange of feldspathic porcelain can be carried out at a temperature as high as 830°C (Jones, 1983), conceivably, an ion-exchange slurry can be developed such that the treatment can be performed at 650°C for a time less than 30 min, followed by rapid cooling to room temperature. The advantage of such a process will be to enhance the flexural strength and narrow the distribution of strength values. Both approaches should be explored further to test the feasibility of these methods and their potential clinical benefits.
ACKNOWLEDGMENT This study was supported in part by Research Grant DE06672 and DE09307 from the National Institute ofDental Research, Bethesda, MD 20892. Received July 19, 1991/Accepted August 30, 1991 Address correspondence and reprint requests to: C. Shen Department of Dental Biomaterials College of Dentistry University of Florida Gainesville, FL, USA 32610-0446
REFERENCES Anusavice KJ, Hojjatie B (1991). Effect of thermal tempering on strength and crack propagation offeldspathic porcelain. J Dent Res 70:1009-1013. Anusavice KJ, Shen C, Gray A, Lee B (1990). Strengthening of feldspathic porcelain by ion exchange and tempering. J Dent Res 69:210, Abstr. No. 816. DeHoff PH, Anusavice KJ (1989). Tempering stresses in feldspathic porcelain. J Dent Res 68:134-138. Hood HP, Stookey SD (1957). Glass articles ofhigh mechanical strength, U.S. Patent 2,779,136. Jones DW (1983). The Strength and Strengthening Mechanisms of Dental Ceramics, In: Dental Ceramics. Proceedings of the First International Symposium on Ceramics. J.W. McLean, Ed., Quintessence Publishing Co., Inc., Chicago, 83-141. Katz S (1989). High strength feldspathic dental porcelains containing crystalline leucite. U.S. Patent 4,798,536. Kirstein AF, Woolley RM (1967). Symmetrical bending ofthin circular elastic plates ofequally spaced point systems. JRes Natl Bur Stds 71(C):1-10. Kistler SS (1962). Stresses in glass produced by nonuniform exchange of monovalent ions. J A m Ceram Soc 45:59-68. McCabe JF, Walls AWG (1986). The treatment of results for tensile bond strength testing. JDent 14:165-168. PiddockV, Qualtrough AJE, Brough I (1991). An investigation of an ion strengthening paste for dental porcelains. Int J Prosthodont 4:132-137. Ritter JE, Bandyopadhyay N, Jakus K (1981). Statistical reproducibility of the dynamic and static fatigue experiments. Ceramic Bulletin 60:798-806. Wachtman JB, Capps W, Mandel J (1972). Biaxial flexure tests of ceramic substrates, J Mater 7:188-194.
Strengthening of porcelain by ion exchange subsequent to thermal tempering K.J. Anusavice, C. Shen, B. Vermost, B. Chow
Department of Dental Biomaterials, College of Dentistry, University of Florida, GainesviUe, FL, USA
Abstract. The purpose of this study was to test the hypothesis that high-leucite content feldspathic porcelain can be strengthened best by thermal tempering followed by ion exchange, compared with either thermal tempering or ion exchange alone. The results of biaxial flexure testing indicate that thermal tempering was the most effective way of strengthening the porcelain among the three methods tested. The combined procedure did not yield the greatest strengthening effect as hypothesized, but one that is only slightly stronger than that produced by ion exchange alone. Weibull analysis indicated that the ion-exchange treatment resulted in a narrower distribution of biaxial flexural strength compared with that of untreated or thermally tempered porcelain. The narrower distribution of biaxial flexural strength is considered desirable for enhancing the longevity of ceramic-based dental restorations.
INTRODUCTION It is well established that the presence of surface flaws reduces the effective strength ofglasses and ceramics. Many attempts have been made to achieve tensile strengths closer to the theoretical value calculated from intrinsic cohesive forces by producing flaw-free glass. Etching of glass and ceramic surfaces reduces the influence of surface flaws, which leads to an increase of the strength in the glass, but is not deemed practical ifa chemically aggressive environment is required. Another approach is to produce a more desirable stress distribution within the surface by a thermal anneal treatment. By far, the most effective method of strengthening the glass is to generate a state of compression in the surface. This can be accomplished either by physical tempering or chemical strengthening. Physical tempering is a process whereby a glass article is heated to a temperature below its softening temperature and then quenched rapidly to room temperature, usually in air, molten salt or oil. During the rapid quenching, the exterior of the glass quickly becomes rigid while the interior cools more slowly and continues to contract after the exterior has become hard. Subsequent interior contraction relative to the surface produces compression in the outer surface while the interior is in a state of tension. The presence of this surface compression has been shown to improve the biaxial flexural strength of a body porcelain by as much as 158% (Anusavice and Hojjatie, 1991). Chemical strengthening is an ion exchange process that alters the chemical composition by introducing larger alkali metal cations in the surface layer of the glass while the bulk
of the glass retains the original composition. After cooling, compressive stress develops in the surface region that has changed chemically. Two different mechanisms have been suggested to explain the development of the compressive stress. Hood and Stookey (1957) exchanged lithium ions for sodium ions in the surface layer, resulting in glass with a surface oflower coefficient ofthermal contraction. A favorable stress profile was thus produced upon cooling. Kistler (1962) replaced sodium ions in the surface with larger potassium ions below the transformation range of the glass; the viscosity was so high in that temperature range that stress induced due to crowding of potassium ions in the places of smaller sodium ions could not be relieved. This also yielded a layer of compressive stress in the surface. A previous study (Anusavice et al., 1990) of the effect of thermal tempering and ion exchange on seven dental porcelains showed that both methods are effective in enhancing the strength of dental porcelains. Thermal tempering produced a deeper layer of compressive stress which yielded a greater increase in flexural strength compared with that associated with the ion exchange process. Ion exchange produces a state of compression to a limited depth from the surface, ranging from 30 to 100 pm, and a smaller degree of improvement in biaxial strength is realized. However, the magnitude of the compressive stresses in the surface region of ion-exchanged specimens was much greater than that ofth ermally tempered specimens. In the development of dental porcelains, various attempts have also been made to improve the bulk strength of the material. Those include the use of a higher expansion metal substrate (metal-ceramic), dispersion strengthening (aluminous core material), and controlled crystallization (Dicor castable ceramic, Dentsply, York, PA, USA). It has been shown that crystalline leucites (K~O"A1203"4SiO2)can precipitate from a feldspathic glass matrix with high potassium content, and function as a reinforcing agent of the porcelain. This approach was put to practice by Katz (1989) and a commercial product named Optec HSP (Generic/Pentron, Wallingford, CT, USA) has since become available and has been shown to exhibit a higher strength compared with other commercial porcelains (Katz, 1989). To encourage precipitation ofleucite from feldspathic glass matrix, the potassium-to-sodium ratio is increased. This composition change raises a question as to how effectively leucite-containing porcelain with a reduced sodium content can be chemically strengthened, or physically strengthened
Dental Materials~May 1992 149
in the presence of a reduced volume fraction of glass matrix and an increased concentration ofleucite crystal. It is possible that a combined procedure of ion exchange and thermal tempering can more effectively strengthen porcelain than either method applied alone. The purpose of this study was to test the hypothesis that, despite the lower sodium content, either tempering or ion exchange can strengthen the high-leucite content porcelain, and that a combined treatment ofthermal tempering followed by ion exchange can strengthen the porcelain to generate a greater level than that achieved by either method alone.
MATERIALS AND METHODS Forty-two disks ofOptec HSP porcelain specimens (shade 3.5), 16 mm in diameter and 2.5 mm thick, were prepared. A cylindrical brass mold, approximately 3.5 mm deep was used to form the porcelain disks specimens prior to firing. After the first firing, the specimens were ground to a thickness of 2 mm starting with 60 grit and finishing with 600 grit abrasive paper and then polished through 1 ~m A1203 abrasive prior to glazing. Each disk was glazed at 1016°C and slowly cooled (SC) to room temperature by terminating power to the furnace. Six disks were used as controls. The remaining disks were divided into six groups of six disks each and subject to one of the following six treatments: (1) thermal tempering by forced convective cooling in air (T) at a pressure of 0.34 MPa after a heat soaking at 850°C for 2 rain, (2) a tempering control treatment (TC) consisting of the same heat soaking treatment as in (1) followed by slow cooling in the furnace, (3) ion exchange (I) at 450°C in the furnace for 30 rain using a slurry of the ion-exchange agent Tuf-Coat (GC International, Tokyo, Japan) applied on one side and then fast cooled in ambient air, (4) an ion exchange control treatment consisting of the same heating and cooling cycle as the ion exchange group without application ofion-exchange agent (IC), (5) a combined tempering and ion exchange treatment (TI) consisting of the same heat soaking and rapid cooling cycle used for the tempered group and followed by the same treatment as the ion-exchange specimens, and (6) a tempering-ion exchange control treatment (TIC) produced by combined tempering and ionexchange procedures but without the application of ionexchange agent. Ion exchange was carried out by applying a thin layer of the ion-exchange slurry to the polished surface of the specimen and then skimmed to a uniform thickness with a knife blade while the specimen was stabilized in the trimming fixture at a fixed depth of approximately 250 lam below the fLxture surface. Each of the slurry-coated discs was then placed on a tray to dry for 20 min at 150°C in a small oven. After drying, specimens were heat-treated at 450°C for 30 min in a Unitek burnout furnace and then fast cooled in air. The residue was brushed and rinsed off under running water. All seven groups of specimens were stored in deionized water at 37°C for 7 d prior to testing and analysis. Biaxial flexural strength was determined for six discs per group using a pin-on-three-ball f~xture (Wachtman et al., 1972). The specimens were loaded to failure under water (37°C) at a crosshead speed of 0.5 mm/min using a universal testing machine (Instron Model 1125, Instron Corp. Canton, MA, USA). The load was centrally applied using a flat-ended indenter, with the treated or glazed surface of the disk seated on the three-ball support. The biaxial flexural strength data calculated by an equation proposed by Kirstein and Woolley
150 Anusavice et aL/Strengthening of porcelain
(1967) were also fitted by nonlinear regression into the twoparameter Weibull distribution function, which is expressed as follows: Pf = 1 - exp - G/G o (1)
[(
)m]
where Pf is the probability of porcelain failure under a stress, G is the biaxial flexural strength calculated from the failure load, G0is the characteristic strength which is for normalizing the stress term and has no physical meaning, and m is the shape factor. The calculated biaxial flexural strength values of the test group were first sorted in ascending order and a value of probability of failure (< 1.0) was then assigned to each strength in the order of 1/(n+ 1) to n/(n+l), where n is the total number of specimens. Since there were six specimens for each group, the probability of failure was 14.3% for the specimens with lowest strength and 85.7% for those with highest strength. The values of Goand m ofeach group were determined by fitting the biaxial flexural strength and the assigned probability of failure into Eq. (1) by a nonlinear regression procedure. RESULTS The mean biaxial flexural strength values and their 95% confidenceintervals for the seven treatment groups are shown in Table 1. The improvement of strength ranges from a negative 5.2% for the ion exchange control to a positive 43.5% for tempered specimens. Analysis of variance indicated that a significant difference (p < 0.01) existed among groups. Results of Tukey's HSD test indicated that there were no statistically significant differences (p > 0.05) among the four control groups, between groups T and TI, and between TI and I. The results of Weibull analysis are shown in Table 2. The value of m, which is shown with its respective 95% confidence interval, indicates the distribution ofbiaxial flexural strength TABLE 1: BIAXlAL FLEXURALSTRENGTHAS A FUNCTIONOF TREATMENT Treatment
Code
Strength(aPa)
Slow Cooled Tempered Control Tempered Ion Exchanged Control Ion Exchanged
SC TC T IC I
143.2 + 142.5 + 205.5 + 135.8 + 168.9 +
Tempered & Ion
TIC
152.8 + 21.8
13.8 11.7 15.5 7.3 16.9
% Change vs. SO 0.0 -0.5 I
I
43.5 -5.2 18.0 6.7
Exchanged Control Tempered & Ion Exchanged TI 183.2 + 14.6 IJ 27.9 * Vertical lines connect mean values that are not statistically different at p = 0.05.
TABLE 2: RESULTS OF WEIBULL ANALYSIS AS A FUNCTIONOF TREATMENT Characteristic Treatment Code Strength (% Mea) Slow Cooled SC 150.1 Tempered Control TC 147.7 Tempered T 218.5 Ion Exchanged Control C 139.4 Ion Exchanged I 176.9 Tempered & Ion Exchanged TIC 162.9 Control Tempered & Ion Exchanged TI 186.4
Shape Factor
Goodnessof-Fit
(m)
( °/o )
10.8 _+6.7 10.7 + 3.1 8.7 + 6.4 17.4 +11.0 14.3 + 7.4 7.2 + 2.2
91.1 97.4 89.8 90.9 94.5 97.6
14.6 + 9.9
91.4
100.9:~
,' // 4 :3/d
Slow Cooled
:'
0.8
• TemperedContro~
0.7
:'
i
Ion Exchanged Control
"6
~
/~-~
,': /~/
Ion ExchangedContro,
0.6-
I/ //
!
Tempered& _.
values obtained experimentally. Fig. 1 shows the experimental data along with best-fit Weibull distribution for the four controls. Fig. 2 shows the experimental data and Weibull distribution of three treated specimens compared with that for the slow-cooled group. Fig. 3 shows the experimental data and Weibull distribution of steps from slow cooled to the combined procedure. The experimental data showed that the ion exchange process resulted in a narrower distribution of strength values while the tempering process produced a broader distribution of strength values.
: /i
/
/
/
! Oli~%
0.5-
id
; ~E/
o
0.3
th
@
/// Z':"
"~ 0.4.0
0.2
DISCUSSION
0.12Y,:" 0.0-
i
r
i
i
r
50
75
100
125
150
E 175
Biaxial Flexural Strength,
i
i
200
225
MPa
Fig. 1. Effect of cooling cycle on the flexural strength of porcelain with a high-leucite content. 10
/I
09.... 0.8
Slow Cooled • Tempered
0.7
Ion Exchanged
--
Tempered & Ion Exchanged
0.6-
"6
t
---
0.4
:}'
t
/
/
f
/ [ .I
01-
/"
, ." •
/
/
/ I
/
'
,*o
it /
/
02-
o," ,: .'
I
'
.0 0 13- 0,3~
:
I
i,/I
"'"'"
• ,, ,"'"
I
J / I /'
05
/' "
iI
"/"--'--'T / /
, / /'
/
./
iI / /
," "//,/:
/
/ /" //.-"
00-
75
5O
,00
~25
~so
,Ts
2oo
Biaxial Flexural Strength,
22s
250
275
MPa
Fig. 2. Effect of treatment on the flexural strength of porcelain. 10
/
09 - I) 0.8 (1)
B
~
0.7-
-
,! / I / /' / ~1 //~
-- o Tempered & Ion
--
[
Ion Exchanged
.z~
0.4-
/
,,> i/
4/ /
~
"..?
.
,,"" ,'"
:"
/
//
,"
,,""
/~
,' // .
./:"/:
/...":
i? "
0.2-
~./
I
J/ .(/~
~-- 0.3-
:" +,' ,, e."
/
!~/
"6 0.5-
"
/
i iI"
:) Tempered &
..'"-- - -
II / ?/ I '~
[I!
Slow Cox)led
- • Tempered Exchanged Control
0.6-
~" ---/-~7~
,I
//
0.0~
75
,oo
,2s
,50
,75
~0
Biaxial Flexural Strength,
22s
~50
27s
MPa
Fig. 3. Biaxial flexural strength of porcelain as influenced by the sequence of
According to a published patent (Katz, 1989), the leucite content of the Optec porcelain is at least 45 w/o and typically can be within the range of 55 to 75 w/o. Assuming that the composition disclosed in the patent description is the same as the material used in the study, i.e., a porcelain that contains 45 w/o ofleucite, for 100 g ofOptec HSP, the approximate mole ratio of SiO2 : A1203 : K20 : Na20 : Li203 : CaO in glassy phase of the porcelain is 0.57 : 0.06 : 0.05 : 0.06 : 0.05 : 0.02. This analysis shows that there are more lithium and sodium than potassium in the glassy matrix. This chemical makeup may favor a situation in which Na ÷replaces Li÷, while K÷replaces Na ÷duringion exchange treatment. The 18% improvement in the biaxial flexural strength value is low compared with the range of 31.9% to 96.8% reported previously for seven commercial porcelains of higher sodium content (Anusavice et aI., 1990). It has been suggested that the existing K÷within the structure can reduce the driving force for potassium uptake near the surface region and result in a lower percentage of improvement of the failure strength values (Piddock et al., 1991). It is important, nonetheless, to note that even a high leucite containing porcelain can be strengthened by ion exchange as long as smaller monovalent ions are present. The four control groups were heat soaked and then cooled from the following temperatures: 1050 ° C for slow cooling, 850°C for the tempered control, and 450°C for the ion exchange and combined treatment control. It is reasonable to assume that an equilibrium state was being established prior to cooling and the temperature of heat soaking influenced the stress distribution in the porcelain surface upon cooling. Weibull analysis indicated that the group cooled from the lower temperature had a narrower distribution of flexural strength, despite the fact that the mean biaxial flexural strength of the groups was not statistically different (p > 0.05). The ion exchange control group, which was fast-cooled to room temperature from 450°C, exhibited the highest m value. This implies that 450°C is low enough for Optec HSP to avoid development of a nonuniform stress in the surface which is often associated with rapid cooling from a higher temperature. Fig. 2 shows the effect of treatments on the distribution of flexural strength of the porcelain. The tempered group has the highest mean strength value and the widest distribution of the four groups. It indicates that, although tempering is the most effective method of strengthening, the results are less predictable because of a wider distribution. Strengthening by ion exchange is less effective but the distribution of failure strength values is much narrower. This may not appear to be of significance if one compares only the average strength values among groups. In reality, if we assume that a failure rate of I to 5% is considered acceptable, the average strength values shown in Table 1 become meaningless. From the point of designing a prosthesis, the failure strength corresponding to
combined treatment.
Dental Materials~May 1992 151
failure level of 5% or less is more meaningful. Table 3 shows the predicted strength at 1% and 5% of failure by Weibull distribution for all seven groups. At 5% of failure level, the strength values are 155,144 and 152 MPa for tempering, ion exchange, and tempering and ion exchange, respectively. At the 1% failure level, the strength values are 128, 128 and 130 MPa for tempering, ion exchange, and tempering and ion exchange, respectively. It becomes clear that a higher mean value of strength, such as that produced by tempering, is not necessarily equivalent to a higher resistance to fracture when the acceptable failure level is set at 1% or 5%. Fig. 3 shows the changes in the biaxial flexural strength value at each step from slow cooling to tempering and ion exchange. The pattern of changes implies that ion exchange treatment at 450°C for 30 rain was not sufficient to produce an equilibrium stress state, but to cause significant stress relaxation within the tempered layer. Meanwhile, the application of ion-exchange paste to the stress-relaxed disk surfaces yielded a strength value intermediate to those associated with ion exchange and tempering alone. It is interesting to note that the shape factor of the tempered, and tempered and ion exchange control specimens is about the same, while ion exchange improves the shape factor (Table 2). This indicates that the tempering procedure yields a wide distribution of strength values and the ion exchange control cycle can only reduce the magnitude but not the variability of the surface stress within the tempered layer. We may conclude from Figs. 1 and 3 that 450°C is low enough to avoid development of nonuniform stress from coolingbut not high enough to relieve tempered stress completely. The number of specimens used by investigators for a twoparameter analysis of the Weibull distribution is variable. McCabe and Walls (1986) argued that the value ofm gives an indication of the reliability of the flexural strength; therefore, the 95% confidence interval of the m value calculated from experimental data should be as small as possible. They concluded from their study that a minimum of 30 would be needed to perform a meaningful analysis by two-parameter Weibull distribution. In fact, the m values shown in Table 2 are not statistically different for all seven groups by Tukey's HSD because of the relatively wide range of 95% confidence interval of the m value of each group. Since the strength values of the four control groups of the present study were not significantly different (p > 0.05), we can treat the strength values offour control groups as one group of 24 specimens, and fit the value along with respectively assigned probability of failure based on 24 specimens into equation (1). The estimated values of Go and m are 148 MPa and 11.1 _+ 1.2, respectively. The m value calculated from a population of 24 specimens has a smaller 95% confidence interval compared with that of the individual control groups. This result is consistent with the conclusionofMcCabe and Walls (1986). In the case of lifetime predictions, there are more variables involved and a minimum of 100 or more has been suggested by Ritter et al. (1981). Obviously, if the m value obtained is for the purpose of predicting the reliability of a particular material, a higher number of specimens is needed to assure a small range of 95% confidence interval. The m values calculated in this study, which are based on 6 specimens with a goodnessof-fit of 90% or greater, should only be used for the purpose of comparing the effectiveness of treatments on each particular porcelain. Further studies should concentrate on the analysis of the
152 A nusavice et aL/Strengthening of porcelain
temperature and time for which degree of stress relaxation will be least affected while significant strengthening by ion exchange can still be achieved. It is also conceivable that additional strengthening can occur by changing the sequence of treatment to ion exchange followed by tempering. DeHoff and Anusavice (1989) showed by finite element analysis that rapid cooling from 650°C can be effective in strengthening feldspathic porcelain. Since ion exchange of feldspathic porcelain can be carried out at a temperature as high as 830°C (Jones, 1983), conceivably, an ion-exchange slurry can be developed such that the treatment can be performed at 650°C for a time less than 30 min, followed by rapid cooling to room temperature. The advantage of such a process will be to enhance the flexural strength and narrow the distribution of strength values. Both approaches should be explored further to test the feasibility of these methods and their potential clinical benefits.
ACKNOWLEDGMENT This study was supported in part by Research Grant DE06672 and DE09307 from the National Institute ofDental Research, Bethesda, MD 20892. Received July 19, 1991/Accepted August 30, 1991 Address correspondence and reprint requests to: C. Shen Department of Dental Biomaterials College of Dentistry University of Florida Gainesville, FL, USA 32610-0446
REFERENCES Anusavice KJ, Hojjatie B (1991). Effect of thermal tempering on strength and crack propagation offeldspathic porcelain. J Dent Res 70:1009-1013. Anusavice KJ, Shen C, Gray A, Lee B (1990). Strengthening of feldspathic porcelain by ion exchange and tempering. J Dent Res 69:210, Abstr. No. 816. DeHoff PH, Anusavice KJ (1989). Tempering stresses in feldspathic porcelain. J Dent Res 68:134-138. Hood HP, Stookey SD (1957). Glass articles ofhigh mechanical strength, U.S. Patent 2,779,136. Jones DW (1983). The Strength and Strengthening Mechanisms of Dental Ceramics, In: Dental Ceramics. Proceedings of the First International Symposium on Ceramics. J.W. McLean, Ed., Quintessence Publishing Co., Inc., Chicago, 83-141. Katz S (1989). High strength feldspathic dental porcelains containing crystalline leucite. U.S. Patent 4,798,536. Kirstein AF, Woolley RM (1967). Symmetrical bending ofthin circular elastic plates ofequally spaced point systems. JRes Natl Bur Stds 71(C):1-10. Kistler SS (1962). Stresses in glass produced by nonuniform exchange of monovalent ions. J A m Ceram Soc 45:59-68. McCabe JF, Walls AWG (1986). The treatment of results for tensile bond strength testing. JDent 14:165-168. PiddockV, Qualtrough AJE, Brough I (1991). An investigation of an ion strengthening paste for dental porcelains. Int J Prosthodont 4:132-137. Ritter JE, Bandyopadhyay N, Jakus K (1981). Statistical reproducibility of the dynamic and static fatigue experiments. Ceramic Bulletin 60:798-806. Wachtman JB, Capps W, Mandel J (1972). Biaxial flexure tests of ceramic substrates, J Mater 7:188-194.
