- Email: [email protected]
Optimization of glass-ceramic crystallization based on DTA exotherm analysis
Dent Mater 10:167-171, May, 1994
Optimization of glass-ceramic crystallization based on DTA exotherm analysis Douglas E. Parsell 1,2,Kenneth J. Anusavice I
1University of Florida, College of Dentistry, Department of Dental Biomaterials, Gainesville, Florida, USA Wow at: Department of Restorative Dentistry, Biomaterials, University of Mississippi Medical Center, Jackson, Mississippi, USA
ABSTRACT Objectives. Crystallization of glass-ceramics is traditionally achieved through a two-stage heat treatment consisting of an isothermal nucleation stage followed by an isothermal growth stage. A method for determining a more efficient heat treatment schedule for a glass-ceramic material using a thermal analysis technique is proposed. The goal of an optimized heat treatment schedule is the production of a glass-ceramic with a desired microstructure (number of crystals per volume) in the shortest amount of time. Methods. The proposed method involves differential thermal analysis (DTA) to measure glass crystallization exotherm characteristics which are correlated with the population density of growing crystals, and therefore, to the effectiveness of any prior heat treatment. Traditional thermal processing parameters were investigated and optimized. A method for generating a more efficient heat treatment schedule composed of a series of increasing heating rates was also demonstrated. Results. The thermal analysis method measured a significant effect upon the number density of crystals generated as a function of several experimental variables. Micrographs from samples crystallized with a more time-efficient heat treatment schedule were shown to have equivalent crystal number densities compared to those crystallized with a more time consuming, traditional schedule. Significance. This work demonstrated that a rapid thermal analysis method was capable of measuring the relative effectiveness of heat treatment schedules to generate crystalline populations. A novel heat treatment schedule was developed based on progressive adjustment of processing heating rates to generate the maximum crystal population in the shortest amount of time. INTRODUCTION For a glass-ceramic material to serve as an alternative to PFM, the material must exhibit sufficient flexural strength, toughness, and wear resistance while maintaining superior esthetics. Crystal population density is one of the major microstructural parameters which influences these properties (McMillian, 1979). The magnitude of the crystal population density, for a given material processing time, is dependent
upon the efficiency of the heat treatment schedule. Therefore, optimization of glass-ceramic heat treatment schedules is an important step in the development of glass-ceramics for use as dental materials. Thermal processing of glass-ceramic materials has been based on a conventional two-stage heat treatment, namely, nucleation and growth (Stookey, 1958; Kingery et al., 1976). Although two-stage heat treatment schedules have been able to produce the required glass-ceramic microstructures, it is proposed that more refined, multisegmented schedules can produce an equivalent glass-ceramic microstructure with less processing time. In the past, the lack of temperature-controlling equipment capable of executing multisegmented heat treatment programs has limited heat treatments to simple schedules. Since this limitation no longer exists, a heat treatment schedule can be developed which more accurately approximates the optimum time vs. temperature profile for the most rapid development of a desired glass-ceramic microstructure. The aim of an optimum heat treatment schedule is the development of a given nuclei population from their initial size to their final size as rapidly as possible. The growth rate of sufficiently small crystals is a function of crystal size and temperature (Kelton, 1991). For a given nucleus size, there is a unique temperature corresponding to the maximum crystal growth rate. Therefore, as nuclei grow, the temperature at which the maximum growth rate occurs also increases (Kelton, 1991). For small nuclei, the temperature for a maximum growth rate is low, and the growth rate is also low. This corresponds to a more shallow-sloped section on a time vs. temperature heating schedule plot. For larger nuclei, such as those that have been produced from a heterogeneous body, the temperature for maximum growth is high, and the corresponding growth rate is also high. This crystal growth process is represented by a steeply sloped section on a time vs. temperature heating schedule plot. As a result of the influence of crystal size on the temperature for maximum crystal growth rate, the optimum schedule should be a smooth curve of increasing slope as time and temperature increase. Schedules which increase temperature too rapidly will generate temperatures which are above the Dental Materials/May 1994 167
temperature for maximum crystal growth for the given crystal size. As the temperature continues to increase above the maximum growth rate temperature, crystal growth rates will decrease until the temperature for thermodynamic phase stability is reached and the crystal begins dissolving back into the glass matrix (Leontjewa, 1942; Swift, 1947; Morley, 1965; Meiling and Uhlmann, 1967; Wagstaff, 1969). The crystal population density is therefore reduced by heating rates faster than the optimum rate. The purpose of this paper is to demonstrate a method which rapidly generates the most efficient heat treatment schedule for the crystallization of a glass.
MATERIALSANDMETHODS Thermal Analysis. Differential thermal analysis (DTA) and differential scanning calorimetry (DSC) have been used previously to determine the effect of nucleation temperature and nucleation time on the crystallization of several different glass-ceramic systems (Marotta et al., 1981; Ray and Day, 1990; Xu et al., 1991). For this technique, samples are nucleated in situ and then heated through a higher temperature region where crystal growth rates are rapid enough to produce easily detectable exotherms. Both the magnitude of the crystallization peak and the time for the exotherm maximum to occur can be correlated to the number of growing crystals and therefore to the efficiency of the prior nucleation heat treatment. The validity of these techniques has been demonstrated through comparison with microstructural analysis (Parsell, 1992) and through mathematical modelling (Weinberg, 1991). Thermal output data were generated from DTA analysis because the crystallization temperature needed for this material was in the upper limits of the usable temperature range for DSC. The exothermic peak was measured at an isothermal temperature within the crystallization range. The time to reach the exothermic maximum was measured during an isothermal stage to improve the resolution ofthe technique. In measuring the characteristics of an exotherm generated during a heating segment, the accelerating crystallization rate associated with the increasing temperature effectively compresses the exothermic peak along the time axis as crystallization progresses. It is expected that this would decrease the resolution of the measured time to reach the maximum in the exotherm profile. The glass used in this study is of the molar composition 30.5% Li20, 2.5% A1203, 5.0% CaO, 61.0% SiO2, and 1.0% P205. This glass is designated as LACSP. The glass batch was made from powders of Li20-2SiO 2 (Specialty Glass, Oldsmar, FL, USA) and reagent grade A1203,P205 and CaCO 3(Fisher Scientific, Fair Lawn, NJ, USA). The powders were weighed and batched in a glove box under a flowing nitrogen atmosphere. The powder batch was ball-milled for 24 h. The glass was melted in a covered platinum crucible in an air atmosphere for approximately 24 h at 1325°C. The glass was poured at 1350°C into deionized water at 27°C. X-ray diffraction analysis confirmed that no detectable amounts of nucleation occurred during cooling from the melt. The quenched glass was washed, crushed and sieved to the desired particle sizes. Thermal analysis samples were composed of 16.0 mg of glass grains that were distributed uniformly in a platinum pan. Grains sized approximately 900 pm in diameter were used to ensure that the thermal output associated with bulk 168
Parsell&Anusavice/Glass-ceramicheattreatmentoptimization
25O .~E
200
~!~:
!ilii%iiiiiiiiiiii
=E E
150 ~ilili! %~ ..... 0
1o0 LU 0
~
5o 0
16
%~o% 30 e. e e~
Nuclei DenSitY (.~O..4numberI p.m
Fig. 1. Simulated relationship between crystal density, particle size and the time to reach the exothermic maximum.
crystallization could be clearly differentiated from that associated with surface crystallization. Because the primary use of this technique is to rapidly establish a heat treatment schedule, such that further material testing can be performed to evaluate material usefulness, each data point is representative of one thermal analysis run. Previous work with a similar system produced data for the thermal analysis method with an average standard deviation of approximately 1.1 min (Parsell, 1993). Computer Simulation of Crystallization. To better understand the relationships between experimental parameters and the characteristics of the measured crystallization exotherm, a computer model was developed to simulate the crystallization process monitored during thermal analysis. The computer model simulates the exothermic process by calculating the heat produced during crystallization of one of the glass grains within the sample pan. The total thermal output was calculated by multiplying the heat released from this single glass grain by the number of glass grains within the sample pan. The system was geometrically configured as a sphere of glass which crystallizes by surface and bulk crystal growth. Impingements of bulk crystals on bulk crystals and bulk crystals on surface crystals were takeninto consideration. The experimental parameters that can be varied in the model are bulk nuclei density, isothermal temperature, grain size, and time interval between the onset of surface crystallization and the onset of bulk crystallization. The computer model generates the simulated thermal output as a function of time by equating the volume of glass transformed for each iteration of the program to the thermal output for that time segment. The accuracy of this model has been previously demonstrated through comparison with stereology data (Parsell, 1992; 1993). The calculated relationships between bulk nuclei density, isothermal temperature, and time to exotherm maximum are shown graphically in Fig. 1.
26
• 60
800 24-
=
.c_
Time at Nucleation Temperature Data Heating Rate Data
E
"50
E
.E
E
600
22
' 40
&
._o
E
?%
' 30
400 w
._c E
~ IX
20-
o
"G
'20
Z
200
18I.-
leJ 400
~
T
,
.lO
0
o
4so
soo
s'so
eoo
Nucleation Temperature (°C)
Fig. 2. Time to reach exotherm maximum as a function of nucleation temperature for the LACSP system as determined by DTA.
Optimization Method. As previously noted, the form of the optimum heat treatment schedule should be a smooth curve of continuously increasing slope. The thermal analysis technique can be used to derive an approximation of this curve by monitoring changes in the number of stable nuclei per unit volume as the heat treatment is modified. Any modification to the heat treatment schedule which decreases the nuclei population density from the maximum nuclei population density determined from a baseline heating rate schedule represents a deviation from the optimum timers, temperature curve. The baseline heating rate schedule is the most rapid constant heating rate schedule which produces a glass-ceramic microstructure of the desired crystal population density. In the present study, the traditional isothermal nucleation temperature and time were measured. The baseline heat treatment schedule was the isothermal nucleation heat treatment followed by a 0.5°C/min heating rate to 650°C. The sample was fully crystallized with this heat treatment. The crystal population density produced by the baseline heat treatment was measured through optical stereology. Bulk samples were cut cross-sectionally, polished through 1 pm alumina abrasive and etched with 2% HF solution. A series of 6 optical micrographs was taken at randomly selected locations on the sample surface. The crystal population density was calculated by dividing the number of crystals counted by the area sampled. This value was used as the crystal population density which should be maintained while time-saving modifications are made on the heat treatment schedule. Heat treatment processing time is optimally reduced by successive increases in the heating rate which do not result in a significant decrease in the crystal population density from that established from the baseline heat treatment schedule.
lo
1'5
2'0
25
30
35
40
Time to Exotherm Maximum (min)
Fig. 3. Effect of nucleation time and healing rate from the nucleation temperature to the growth temperature on the time to reach the exothermic maximum for LACSP.
The thermal analysis was performed with a Seiko TG/DTA 320 unit (Seiko Industries USA, Torrance, CA, USA). Alpha alumina was used as the reference material. All samples were analyzed in an atmosphere of flowing nitrogen at 200 mL/min. To ensure that the heating rate at which a sample approaches the isothermal temperature did not influence the measured characteristics of the exotherm, all samples were subjected to identical heat treatments in the 10°C temperature interval preceding the isothermal temperature. An isothermal temperature of 625°C was used for the majority of the thermal analysis experiments. To increase the sensitivity of the method, some experiments have used a slightly lower isothermal temperature. The increased sensitivity is predicted by the computer model (Fig. 1), but it should also be noted that too low an isothermal temperature will produce an exotherm too subtle to be readily detected. To test the effectiveness of the experimentally determined heat treatment schedule, LACSP discs (16 mm dia x 3 mm thick) were made. One set of six discs was heat treated with the experimentally derived optimum schedule. The other set of 6 discs was nucleated for 3 h at 505°C and then heated at 0.5°C/min until crystallization was complete at 650°C. Micrographs were made from each set of discs using the standard techniques previously described. Crystal population densities from each set of discs were determined by counting crystals per cross-sectional area. Student t-tests were conducted on the values to compare the differences between the two heat treatment processes (p<0.05). RESULTS
The first set of measurements was designed to determine the most effective nucleation temperature for a conventional twoDental Materials~May 1994 169
32"
800" $
data for 0.5 to 1.0°C/min heatingrate transition data for 1.0 to lO.O°C/min heating rate transition
31 A C
30
600
E
E
29 ¸
G" o
o}
(1) UJ
28'
400
2
o~ CL
E l-
E I-
27'
26 500
s;0
s4o
s6o
580
600
200
620
Temperature of Heating Rate Change (°C)
Fig. 4. Effect of variation in the temperature of a heating rate transition on the time to reach the exothermic maximum for the LACSP system. 0
o
stage heat treatment. A constant nucleation time of 3 h was used. The minima in a plot of nucleation temperature vs. time to exotherm maximum corresponds to the most efficient nucleation temperature. The results shown in Fig. 2 indicate that the most efficient temperature was between 500°C and 510°C. The effect of hold time at the nucleation temperature is shown in Fig. 3. The nucleation temperature used was 505°C. Saturation of the nuclei population occurred in approximately 10 h. Since the heating rate of 10°C/min between the nucleation temperature and the isothermal temperature is quite rapid for this set of experiments, the measured saturation is associated with nuclei originating from fairly stable nucleation sites. From the data in Fig. 3, a nucleation time of 3 h was determined to be sufficient for the development of the bulk glass-ceramic microstructure. To explore the effect of the heating rate on the nuclei density, a series of heating rates ranging from 0.5°C/min to 50°C/min were used between the nucleation stage and the isothermal stage. For all samples, the predetermined nucleation heat treatment of 505°C for 3 h was used. The effect of variation in heating rate on the time to reach the exotherm maximum is also shown in Fig. 3. The second set of experiments was designed to determine the best temperature at which a transition could be made from a heating rate of 0.5°C/min to one of 1.0°C/min. The baseline crystal population density was measured optically as 1.64x10-4/~m2. After the nucleation treatment, samples were heated at 0.5°C/min to a series of temperatures where the heating rate was increased to 1.0°C/rain. Fig. 4 shows the time for a sample to reach the exothermic maximum as a function of the temperature of transition from one heating rate (0.5°C/ rain) to a more rapid heating rate (1.0°C/min). It can be seen that the most efficient temperature from which to increase the heating rate is 550°C. This is the lowest temperature at which the heating rate can be accelerated without a significant 170 Parsell&Anusavice/Glass-ceramicheattreatmentoptimization
400
Time (min)
~.5.
DTA-derived optimum processing schedule for the LACSP system.
decrease in the crystal population density. Using a schedule with the transition temperature determined from Fig. 4, the second set of data in Fig. 4 shows that the best temperature to initiate the transition from 1.0°C/rain and 10°C/min is 600°C. To increase the sensitivity of the method, a lower isothermal temperature was used for the 1.0°C/min to 10.0°C/min transition data as compared to 0.5°C/min to 1.0°C/min transition data. Combining all of the previous results produces a good approximation of the optimum time v s . temperature schedule for the LACSP system (Fig. 5). The microstructure of the LACSP materials crystallized with the heating schedule shown in Fig. 5 appeared to be composed of crystals of two distinct sizes. The typical material microstructure is seen in Fig. 6. X-ray diffraction revealed both crystal structures to be lithium disilicate. It is possible that the small crystals are cross-sectional views of dendritic arms from the larger spherulites. The stereological data indicated that the thermal analysis method was effective in producing a material of nearly equivalent crystal density to that achieved using the slow heating rate schedule. For the optimized schedule, the average diameter of the small crystals was 15 pm _+ 2.1 ~m, while the larger spherulites had an average diameter of 91 pm _+5.4 pro. For the slow heating rate samples, the average diameter of the small class of crystals was 71 pm _+ 2.6 ~m, while the larger spherulites had an average diameter of 102 ];m _+5.3 pro. Although the difference in size of the small crystals was significant (p<0.05), there was no difference between the average diameters of the large spherulites.
ACKNOWLEDGEMENT This study was supported by NIH-NIDR Grant DE09307.
Received March 1,1994/AcceptedApril 29, 1994 Address correspondence and reprint requests to: Douglas E. Parsell Department of Restorative Dentistry, Biomaterials University of Mississippi Medical Center School of Dentistry Jackson, MS, USA 39216
REFERENCES Fig. 6. Optical micrograph (400x) of etched LACSP that has been fully crystallized using the optimum heat treatment schedule,
DISCUSSION From a standpoint of process engineering, the usefulness of the proposed thermal analysis technique is derived from the reduction in processing time needed to obtain a given microstructure. For the example given above, use of the optimized schedule results in a processing time savings of 3.4 h (37% of total processing time) compared with processing time for the 0.5°C/min schedule. The optimum time vs. temperature schedule has the potential to yield information on the relationships between temperature, crystalline growth rate, and crystal size. The true optimum heat treatment path is constructed from the set of temperatures associated with the maximum growth rate for each crystal size. The time spent at each temperature is a function of the growth rate at the temperature. Therefore, the optimum heating rate at any temperature is also a function of the growth rate at that temperature. The form of the optimum heat treatment schedule generated via the DTA method is approximately exponential. The mechanistic significance of this observation is a possible area of future research. In summary, a thermal analysis technique has been established that characterizes the crystal density in a glass-ceramic material. This was achieved through experimental measurements and also through the use of a computer model. Furthermore, the heating rate between the nucleation region and the growth region was found to have a strong influence on the final crystal density. Of significance is the development ofa method to determine the optimum progression of heating rates to minimize processing time while maintaining the maximum crystal density. The usefulness of this method was confirmed by experimental data. The method should be helpful for rapidly generating effectiveheat treatment schedules for glassceramics. The effectiveness of the heat treatment schedule facilitates the accurate evaluation of the material's properties as well as economic time savings during processing.
Kelton KF (1991). Crystal nucleation in liquids and glasses. Solid State Physics 45:75-177. KingeryWD, BowenHK, Uhlmann DR(1976). Introduction to Ceramics. New York: John Wiley and Sons, 368-374. Leontjewa A (1942). The linear speed of crystallization of potassium, sodium and lithium disilicates. Acta Physicochimica USSR 16:97-101. Marotta A, Buri A, Branda F (1981). Nucleation in glass and differential thermal analysis. J Mat Sci 16:341-344. McMillian PW (1979). Glass-ceramics. New York: Academic Press, 155-244. Meiling GS, Uhlmann DR (1967). Crystallization and melting kinetics of sodium disilicate. Phys Chem Glass 8:62-68. Morley JG (1965). Crystallization kinetics in some silicate glasses. Part 1. Apparatus for the direct measurement of crystal growth at high temperatures. Glass Tech 6:69-76. Parsell DE (1993). Optimization of glass-ceramic heat treatment schedules. Ph.D. Thesis, University of Florida, Gainesville, FL, 54-72. Parsell DE (1992). Optimized nucleation of lithia-based glass-ceramic systems. JDent Res 71:208, Abst. No. 820. Ray CS, Day DE (1990). Determining the nucleation rate curve for lithium disilicate glass by differential thermal analysis. J A m Cer Soc 73:439-442. Stookey SD (1958). Nucleation. In: Kingery WD, editor. Ceramic Fabrication Processes. New York: John Wiley and Sons, 189-195. Swift HR (1947). Effect of magnesia and alumina on rate of crystal growth in some soda-lime-silica glasses. J A m Cer Soc 30:170-174. Wagstaff FE (1969). Crystallization and melting kinetics of cristobalite. J A m Cer Soc 52:650-54. Weinberg MC (1991). Interpretation of DTA experiments used for crystal nucleation rate determinations. J Am Cer Soc 74:1905-1909. XuXJ, Ray CS, Day DE (1991). Nucleation and crystallization ofNa20-2CaO-3SiO 2glass by differential thermal analysis. J A m Cer Soc 74:909-914.
Dental Materials/May 1994 171
Optimization of glass-ceramic crystallization based on DTA exotherm analysis Douglas E. Parsell 1,2,Kenneth J. Anusavice I
1University of Florida, College of Dentistry, Department of Dental Biomaterials, Gainesville, Florida, USA Wow at: Department of Restorative Dentistry, Biomaterials, University of Mississippi Medical Center, Jackson, Mississippi, USA
ABSTRACT Objectives. Crystallization of glass-ceramics is traditionally achieved through a two-stage heat treatment consisting of an isothermal nucleation stage followed by an isothermal growth stage. A method for determining a more efficient heat treatment schedule for a glass-ceramic material using a thermal analysis technique is proposed. The goal of an optimized heat treatment schedule is the production of a glass-ceramic with a desired microstructure (number of crystals per volume) in the shortest amount of time. Methods. The proposed method involves differential thermal analysis (DTA) to measure glass crystallization exotherm characteristics which are correlated with the population density of growing crystals, and therefore, to the effectiveness of any prior heat treatment. Traditional thermal processing parameters were investigated and optimized. A method for generating a more efficient heat treatment schedule composed of a series of increasing heating rates was also demonstrated. Results. The thermal analysis method measured a significant effect upon the number density of crystals generated as a function of several experimental variables. Micrographs from samples crystallized with a more time-efficient heat treatment schedule were shown to have equivalent crystal number densities compared to those crystallized with a more time consuming, traditional schedule. Significance. This work demonstrated that a rapid thermal analysis method was capable of measuring the relative effectiveness of heat treatment schedules to generate crystalline populations. A novel heat treatment schedule was developed based on progressive adjustment of processing heating rates to generate the maximum crystal population in the shortest amount of time. INTRODUCTION For a glass-ceramic material to serve as an alternative to PFM, the material must exhibit sufficient flexural strength, toughness, and wear resistance while maintaining superior esthetics. Crystal population density is one of the major microstructural parameters which influences these properties (McMillian, 1979). The magnitude of the crystal population density, for a given material processing time, is dependent
upon the efficiency of the heat treatment schedule. Therefore, optimization of glass-ceramic heat treatment schedules is an important step in the development of glass-ceramics for use as dental materials. Thermal processing of glass-ceramic materials has been based on a conventional two-stage heat treatment, namely, nucleation and growth (Stookey, 1958; Kingery et al., 1976). Although two-stage heat treatment schedules have been able to produce the required glass-ceramic microstructures, it is proposed that more refined, multisegmented schedules can produce an equivalent glass-ceramic microstructure with less processing time. In the past, the lack of temperature-controlling equipment capable of executing multisegmented heat treatment programs has limited heat treatments to simple schedules. Since this limitation no longer exists, a heat treatment schedule can be developed which more accurately approximates the optimum time vs. temperature profile for the most rapid development of a desired glass-ceramic microstructure. The aim of an optimum heat treatment schedule is the development of a given nuclei population from their initial size to their final size as rapidly as possible. The growth rate of sufficiently small crystals is a function of crystal size and temperature (Kelton, 1991). For a given nucleus size, there is a unique temperature corresponding to the maximum crystal growth rate. Therefore, as nuclei grow, the temperature at which the maximum growth rate occurs also increases (Kelton, 1991). For small nuclei, the temperature for a maximum growth rate is low, and the growth rate is also low. This corresponds to a more shallow-sloped section on a time vs. temperature heating schedule plot. For larger nuclei, such as those that have been produced from a heterogeneous body, the temperature for maximum growth is high, and the corresponding growth rate is also high. This crystal growth process is represented by a steeply sloped section on a time vs. temperature heating schedule plot. As a result of the influence of crystal size on the temperature for maximum crystal growth rate, the optimum schedule should be a smooth curve of increasing slope as time and temperature increase. Schedules which increase temperature too rapidly will generate temperatures which are above the Dental Materials/May 1994 167
temperature for maximum crystal growth for the given crystal size. As the temperature continues to increase above the maximum growth rate temperature, crystal growth rates will decrease until the temperature for thermodynamic phase stability is reached and the crystal begins dissolving back into the glass matrix (Leontjewa, 1942; Swift, 1947; Morley, 1965; Meiling and Uhlmann, 1967; Wagstaff, 1969). The crystal population density is therefore reduced by heating rates faster than the optimum rate. The purpose of this paper is to demonstrate a method which rapidly generates the most efficient heat treatment schedule for the crystallization of a glass.
MATERIALSANDMETHODS Thermal Analysis. Differential thermal analysis (DTA) and differential scanning calorimetry (DSC) have been used previously to determine the effect of nucleation temperature and nucleation time on the crystallization of several different glass-ceramic systems (Marotta et al., 1981; Ray and Day, 1990; Xu et al., 1991). For this technique, samples are nucleated in situ and then heated through a higher temperature region where crystal growth rates are rapid enough to produce easily detectable exotherms. Both the magnitude of the crystallization peak and the time for the exotherm maximum to occur can be correlated to the number of growing crystals and therefore to the efficiency of the prior nucleation heat treatment. The validity of these techniques has been demonstrated through comparison with microstructural analysis (Parsell, 1992) and through mathematical modelling (Weinberg, 1991). Thermal output data were generated from DTA analysis because the crystallization temperature needed for this material was in the upper limits of the usable temperature range for DSC. The exothermic peak was measured at an isothermal temperature within the crystallization range. The time to reach the exothermic maximum was measured during an isothermal stage to improve the resolution ofthe technique. In measuring the characteristics of an exotherm generated during a heating segment, the accelerating crystallization rate associated with the increasing temperature effectively compresses the exothermic peak along the time axis as crystallization progresses. It is expected that this would decrease the resolution of the measured time to reach the maximum in the exotherm profile. The glass used in this study is of the molar composition 30.5% Li20, 2.5% A1203, 5.0% CaO, 61.0% SiO2, and 1.0% P205. This glass is designated as LACSP. The glass batch was made from powders of Li20-2SiO 2 (Specialty Glass, Oldsmar, FL, USA) and reagent grade A1203,P205 and CaCO 3(Fisher Scientific, Fair Lawn, NJ, USA). The powders were weighed and batched in a glove box under a flowing nitrogen atmosphere. The powder batch was ball-milled for 24 h. The glass was melted in a covered platinum crucible in an air atmosphere for approximately 24 h at 1325°C. The glass was poured at 1350°C into deionized water at 27°C. X-ray diffraction analysis confirmed that no detectable amounts of nucleation occurred during cooling from the melt. The quenched glass was washed, crushed and sieved to the desired particle sizes. Thermal analysis samples were composed of 16.0 mg of glass grains that were distributed uniformly in a platinum pan. Grains sized approximately 900 pm in diameter were used to ensure that the thermal output associated with bulk 168
Parsell&Anusavice/Glass-ceramicheattreatmentoptimization
25O .~E
200
~!~:
!ilii%iiiiiiiiiiii
=E E
150 ~ilili! %~ ..... 0
1o0 LU 0
~
5o 0
16
%~o% 30 e. e e~
Nuclei DenSitY (.~O..4numberI p.m
Fig. 1. Simulated relationship between crystal density, particle size and the time to reach the exothermic maximum.
crystallization could be clearly differentiated from that associated with surface crystallization. Because the primary use of this technique is to rapidly establish a heat treatment schedule, such that further material testing can be performed to evaluate material usefulness, each data point is representative of one thermal analysis run. Previous work with a similar system produced data for the thermal analysis method with an average standard deviation of approximately 1.1 min (Parsell, 1993). Computer Simulation of Crystallization. To better understand the relationships between experimental parameters and the characteristics of the measured crystallization exotherm, a computer model was developed to simulate the crystallization process monitored during thermal analysis. The computer model simulates the exothermic process by calculating the heat produced during crystallization of one of the glass grains within the sample pan. The total thermal output was calculated by multiplying the heat released from this single glass grain by the number of glass grains within the sample pan. The system was geometrically configured as a sphere of glass which crystallizes by surface and bulk crystal growth. Impingements of bulk crystals on bulk crystals and bulk crystals on surface crystals were takeninto consideration. The experimental parameters that can be varied in the model are bulk nuclei density, isothermal temperature, grain size, and time interval between the onset of surface crystallization and the onset of bulk crystallization. The computer model generates the simulated thermal output as a function of time by equating the volume of glass transformed for each iteration of the program to the thermal output for that time segment. The accuracy of this model has been previously demonstrated through comparison with stereology data (Parsell, 1992; 1993). The calculated relationships between bulk nuclei density, isothermal temperature, and time to exotherm maximum are shown graphically in Fig. 1.
26
• 60
800 24-
=
.c_
Time at Nucleation Temperature Data Heating Rate Data
E
"50
E
.E
E
600
22
' 40
&
._o
E
?%
' 30
400 w
._c E
~ IX
20-
o
"G
'20
Z
200
18I.-
leJ 400
~
T
,
.lO
0
o
4so
soo
s'so
eoo
Nucleation Temperature (°C)
Fig. 2. Time to reach exotherm maximum as a function of nucleation temperature for the LACSP system as determined by DTA.
Optimization Method. As previously noted, the form of the optimum heat treatment schedule should be a smooth curve of continuously increasing slope. The thermal analysis technique can be used to derive an approximation of this curve by monitoring changes in the number of stable nuclei per unit volume as the heat treatment is modified. Any modification to the heat treatment schedule which decreases the nuclei population density from the maximum nuclei population density determined from a baseline heating rate schedule represents a deviation from the optimum timers, temperature curve. The baseline heating rate schedule is the most rapid constant heating rate schedule which produces a glass-ceramic microstructure of the desired crystal population density. In the present study, the traditional isothermal nucleation temperature and time were measured. The baseline heat treatment schedule was the isothermal nucleation heat treatment followed by a 0.5°C/min heating rate to 650°C. The sample was fully crystallized with this heat treatment. The crystal population density produced by the baseline heat treatment was measured through optical stereology. Bulk samples were cut cross-sectionally, polished through 1 pm alumina abrasive and etched with 2% HF solution. A series of 6 optical micrographs was taken at randomly selected locations on the sample surface. The crystal population density was calculated by dividing the number of crystals counted by the area sampled. This value was used as the crystal population density which should be maintained while time-saving modifications are made on the heat treatment schedule. Heat treatment processing time is optimally reduced by successive increases in the heating rate which do not result in a significant decrease in the crystal population density from that established from the baseline heat treatment schedule.
lo
1'5
2'0
25
30
35
40
Time to Exotherm Maximum (min)
Fig. 3. Effect of nucleation time and healing rate from the nucleation temperature to the growth temperature on the time to reach the exothermic maximum for LACSP.
The thermal analysis was performed with a Seiko TG/DTA 320 unit (Seiko Industries USA, Torrance, CA, USA). Alpha alumina was used as the reference material. All samples were analyzed in an atmosphere of flowing nitrogen at 200 mL/min. To ensure that the heating rate at which a sample approaches the isothermal temperature did not influence the measured characteristics of the exotherm, all samples were subjected to identical heat treatments in the 10°C temperature interval preceding the isothermal temperature. An isothermal temperature of 625°C was used for the majority of the thermal analysis experiments. To increase the sensitivity of the method, some experiments have used a slightly lower isothermal temperature. The increased sensitivity is predicted by the computer model (Fig. 1), but it should also be noted that too low an isothermal temperature will produce an exotherm too subtle to be readily detected. To test the effectiveness of the experimentally determined heat treatment schedule, LACSP discs (16 mm dia x 3 mm thick) were made. One set of six discs was heat treated with the experimentally derived optimum schedule. The other set of 6 discs was nucleated for 3 h at 505°C and then heated at 0.5°C/min until crystallization was complete at 650°C. Micrographs were made from each set of discs using the standard techniques previously described. Crystal population densities from each set of discs were determined by counting crystals per cross-sectional area. Student t-tests were conducted on the values to compare the differences between the two heat treatment processes (p<0.05). RESULTS
The first set of measurements was designed to determine the most effective nucleation temperature for a conventional twoDental Materials~May 1994 169
32"
800" $
data for 0.5 to 1.0°C/min heatingrate transition data for 1.0 to lO.O°C/min heating rate transition
31 A C
30
600
E
E
29 ¸
G" o
o}
(1) UJ
28'
400
2
o~ CL
E l-
E I-
27'
26 500
s;0
s4o
s6o
580
600
200
620
Temperature of Heating Rate Change (°C)
Fig. 4. Effect of variation in the temperature of a heating rate transition on the time to reach the exothermic maximum for the LACSP system. 0
o
stage heat treatment. A constant nucleation time of 3 h was used. The minima in a plot of nucleation temperature vs. time to exotherm maximum corresponds to the most efficient nucleation temperature. The results shown in Fig. 2 indicate that the most efficient temperature was between 500°C and 510°C. The effect of hold time at the nucleation temperature is shown in Fig. 3. The nucleation temperature used was 505°C. Saturation of the nuclei population occurred in approximately 10 h. Since the heating rate of 10°C/min between the nucleation temperature and the isothermal temperature is quite rapid for this set of experiments, the measured saturation is associated with nuclei originating from fairly stable nucleation sites. From the data in Fig. 3, a nucleation time of 3 h was determined to be sufficient for the development of the bulk glass-ceramic microstructure. To explore the effect of the heating rate on the nuclei density, a series of heating rates ranging from 0.5°C/min to 50°C/min were used between the nucleation stage and the isothermal stage. For all samples, the predetermined nucleation heat treatment of 505°C for 3 h was used. The effect of variation in heating rate on the time to reach the exotherm maximum is also shown in Fig. 3. The second set of experiments was designed to determine the best temperature at which a transition could be made from a heating rate of 0.5°C/min to one of 1.0°C/min. The baseline crystal population density was measured optically as 1.64x10-4/~m2. After the nucleation treatment, samples were heated at 0.5°C/min to a series of temperatures where the heating rate was increased to 1.0°C/rain. Fig. 4 shows the time for a sample to reach the exothermic maximum as a function of the temperature of transition from one heating rate (0.5°C/ rain) to a more rapid heating rate (1.0°C/min). It can be seen that the most efficient temperature from which to increase the heating rate is 550°C. This is the lowest temperature at which the heating rate can be accelerated without a significant 170 Parsell&Anusavice/Glass-ceramicheattreatmentoptimization
400
Time (min)
~.5.
DTA-derived optimum processing schedule for the LACSP system.
decrease in the crystal population density. Using a schedule with the transition temperature determined from Fig. 4, the second set of data in Fig. 4 shows that the best temperature to initiate the transition from 1.0°C/rain and 10°C/min is 600°C. To increase the sensitivity of the method, a lower isothermal temperature was used for the 1.0°C/min to 10.0°C/min transition data as compared to 0.5°C/min to 1.0°C/min transition data. Combining all of the previous results produces a good approximation of the optimum time v s . temperature schedule for the LACSP system (Fig. 5). The microstructure of the LACSP materials crystallized with the heating schedule shown in Fig. 5 appeared to be composed of crystals of two distinct sizes. The typical material microstructure is seen in Fig. 6. X-ray diffraction revealed both crystal structures to be lithium disilicate. It is possible that the small crystals are cross-sectional views of dendritic arms from the larger spherulites. The stereological data indicated that the thermal analysis method was effective in producing a material of nearly equivalent crystal density to that achieved using the slow heating rate schedule. For the optimized schedule, the average diameter of the small crystals was 15 pm _+ 2.1 ~m, while the larger spherulites had an average diameter of 91 pm _+5.4 pro. For the slow heating rate samples, the average diameter of the small class of crystals was 71 pm _+ 2.6 ~m, while the larger spherulites had an average diameter of 102 ];m _+5.3 pro. Although the difference in size of the small crystals was significant (p<0.05), there was no difference between the average diameters of the large spherulites.
ACKNOWLEDGEMENT This study was supported by NIH-NIDR Grant DE09307.
Received March 1,1994/AcceptedApril 29, 1994 Address correspondence and reprint requests to: Douglas E. Parsell Department of Restorative Dentistry, Biomaterials University of Mississippi Medical Center School of Dentistry Jackson, MS, USA 39216
REFERENCES Fig. 6. Optical micrograph (400x) of etched LACSP that has been fully crystallized using the optimum heat treatment schedule,
DISCUSSION From a standpoint of process engineering, the usefulness of the proposed thermal analysis technique is derived from the reduction in processing time needed to obtain a given microstructure. For the example given above, use of the optimized schedule results in a processing time savings of 3.4 h (37% of total processing time) compared with processing time for the 0.5°C/min schedule. The optimum time vs. temperature schedule has the potential to yield information on the relationships between temperature, crystalline growth rate, and crystal size. The true optimum heat treatment path is constructed from the set of temperatures associated with the maximum growth rate for each crystal size. The time spent at each temperature is a function of the growth rate at the temperature. Therefore, the optimum heating rate at any temperature is also a function of the growth rate at that temperature. The form of the optimum heat treatment schedule generated via the DTA method is approximately exponential. The mechanistic significance of this observation is a possible area of future research. In summary, a thermal analysis technique has been established that characterizes the crystal density in a glass-ceramic material. This was achieved through experimental measurements and also through the use of a computer model. Furthermore, the heating rate between the nucleation region and the growth region was found to have a strong influence on the final crystal density. Of significance is the development ofa method to determine the optimum progression of heating rates to minimize processing time while maintaining the maximum crystal density. The usefulness of this method was confirmed by experimental data. The method should be helpful for rapidly generating effectiveheat treatment schedules for glassceramics. The effectiveness of the heat treatment schedule facilitates the accurate evaluation of the material's properties as well as economic time savings during processing.
Kelton KF (1991). Crystal nucleation in liquids and glasses. Solid State Physics 45:75-177. KingeryWD, BowenHK, Uhlmann DR(1976). Introduction to Ceramics. New York: John Wiley and Sons, 368-374. Leontjewa A (1942). The linear speed of crystallization of potassium, sodium and lithium disilicates. Acta Physicochimica USSR 16:97-101. Marotta A, Buri A, Branda F (1981). Nucleation in glass and differential thermal analysis. J Mat Sci 16:341-344. McMillian PW (1979). Glass-ceramics. New York: Academic Press, 155-244. Meiling GS, Uhlmann DR (1967). Crystallization and melting kinetics of sodium disilicate. Phys Chem Glass 8:62-68. Morley JG (1965). Crystallization kinetics in some silicate glasses. Part 1. Apparatus for the direct measurement of crystal growth at high temperatures. Glass Tech 6:69-76. Parsell DE (1993). Optimization of glass-ceramic heat treatment schedules. Ph.D. Thesis, University of Florida, Gainesville, FL, 54-72. Parsell DE (1992). Optimized nucleation of lithia-based glass-ceramic systems. JDent Res 71:208, Abst. No. 820. Ray CS, Day DE (1990). Determining the nucleation rate curve for lithium disilicate glass by differential thermal analysis. J A m Cer Soc 73:439-442. Stookey SD (1958). Nucleation. In: Kingery WD, editor. Ceramic Fabrication Processes. New York: John Wiley and Sons, 189-195. Swift HR (1947). Effect of magnesia and alumina on rate of crystal growth in some soda-lime-silica glasses. J A m Cer Soc 30:170-174. Wagstaff FE (1969). Crystallization and melting kinetics of cristobalite. J A m Cer Soc 52:650-54. Weinberg MC (1991). Interpretation of DTA experiments used for crystal nucleation rate determinations. J Am Cer Soc 74:1905-1909. XuXJ, Ray CS, Day DE (1991). Nucleation and crystallization ofNa20-2CaO-3SiO 2glass by differential thermal analysis. J A m Cer Soc 74:909-914.
Dental Materials/May 1994 171