Low Thermal Expansion Sintered LZSA Glass-Ceramics

the

glass researcher Solar POW er Advances in conversion and production efficiencies make future even brighter Peter Wray

With fuel prices approaching $4.50 a gallon, there naturally is more and more interest about the current state of alternative-energy technologies. The jury is still out about the sustainability of ethanol (at least in the U.S.), fuel cells are starting to catch on, but solar power is still the renewable energy source that consumers and planners tend to focus on the most. The good news is that the number of solar technologies continues to grow and solar efficiency records are being broken on regular basis. Across the ideological spectrum in the U.S., solar interest is extremely strong. A poll commissioned by the Solar Energy Industries Association (seia.org) and Schott Solar (schott.com) indicates that 94 percent of Americans believe it’s important for the nation to develop and use solar power and 77 percent think it should be a major priority for the federal government. There are good reasons for the public to be optimistic about solar energy. Photovoltaics and concentrating solar power units move increasingly to cost parity with traditional power sources, and in some isolated cases may already be even. Things are moving fast enough that Co-op America and research group Clean Edge are asserting that the broadbased parity with coal, natural gas will be reached around 2015 American Ceramic Society Bulletin, Vol. 87, No. 7

and will be providing 10 percent of the U.S. energy portfolio by 2025. The firm is confidently predicting that the installed solar PV prices will decline from an average $5.50-$7.00 peak watt (15-32 cents kWh) today to $3.02-$3.82 peak watt (8-18 cents kWh) in 2015 to $1.43-$1.82 peak watt (4-8 cents kWh) by 2025. What’s behind this confidence? Three things: Better sci31

Solar Power

percent efficiency goal. A very different concentrator system set a new net conversion efficiency (i.e., net of auxiliary loads such as water pumps and tracking motors) record of 31.25 percent in February 2008. Researchers at SRNL’s Solar Thermal Test Facility used solar dishes to focus light on a solar receiver that transfers the heat to a Stirling engine. They gained two points of energy conversion efficiency over their old record by improving the optics of the lowiron glass reflectors. A third concentratortype application claims to be hitting high efficiency levels … and high skeptiTandem cells, also called multijunction cells, are individual cells with different bandgaps stacked cism. As reported on in the on top of one another. June issue of The Bulletin, California-based Sungri ence, better applications and better manufacturing. And, the says it can deliver 37.5 percent efficiency using a set of lenses pipeline for even more solar innovations looks very full. to focus sunlight on PV panels backed by a heat sink cooling On the science side, “multijunction concentrator” (also system. Questions remain about Sungri’s methods because known as tandem cells) technology is the leader of the pack. of the company’s resistance to provide any details about the Although still mainly confined to the lab, this technology set cooling system or proposed manufacturing techniques. a world’s record by reaching the 42.8 percent efficiency mark, But, real verifiable and steady progress records are being compared to 15 percent for a typical PV cell. made with thin-film solar cells. In March this year, researchThe MC approach is to stack layers of photosensitive ers at the DOE’s National Renewable Energy Lab created a materials to capture more of the solar spectrum. Each layer CIGS cell that hit the 19.9 percent efficiency mark, on par tuned by the use of a specific mix of materials such as gal- with multicrystalline silicon cells that topped out at 20.3 perlium arsenide, gallium indium phosphide and germanium to cent several years ago. The thin-films have clear application capture a limited band of light. advantages over many solar technologies including weight One continuing testament of the reliability of MC power and flexibility. is the ongoing generation for the Mars Rover missions. Scientists who work with silicon have been hard at work, The efficiency record was hit through the work of a consortium that included DuPont, and the University of Delaware, as part of a $100 million Very High Efficiency Solar Cell program funded by the Defense Advanced Research Projects Agency. The consortium’s goal is to reach 50 percent efficiency by 2010 with portable cells that can be produced for $1,000 or less. The VHESC method is a variation of the previous piggybacking methods. It splits up the light spectrum and then concentrates it on the appropriately tuned cell. The record was set with a five-layer system, and research say they expect the addition of a sixth layer will allow them to reach their 50 32

too, and one of the promising techniques takes its cue from the tandem cells approach. Researchers working on a joint project for Ersol Thin Film GmbH and Schott Solar plan to combine amorphous and microcrystalline techniques in a two-layer cell that they hope will capture the entire light spectrum and gain a 50 percent increase over ordinary multicrystalline cells.

More POW emerging The world is on the brink of a new wave of solar technologies, and, as in other fields, nanotechnologies will play the starring role of matchmaker. American Ceramic Society Bulletin, Vol. 87, No. 7

Exciton fission for an ultra-high efficiency, low cost solar cell.

Penn State is continuing to experiment with a next generation solar power approach by growing silicon nanowires to create high aspect ratio semiconductor heterojunction solar cells.

For example, DOE has given an MIT research team nearly $1 million to continue its work on a new type of MC approach using Cd or Pb semiconducting nanocrystal quantum dots. The DOE is providing a similar grant to Rochester Institute of Technology for another MC method using quantum dots incorporated into a gallium arsenic cell to improve IR absorption in hopes of achieving a 40 percent efficiency level. DOE money is also going to Penn State for a photo cell that will use silicon nanowires grown on a silicon oxide-glass substrate and to Solasta Inc. for a carbon nanotubes PV system that will provide more efficient pathways for photons and charge carriers.

The trail to energy parity doesn’t only go through the lab. Manufacturing costs are rapidly dropping, too, and those improvements are worthy of a separate story. But, one example is worth noting: Nanosolar, using a proprietary nanoparticle ink, has raised the production bar with a $1.65 million machine that can print thin-film solar cells at the rate of 100 feet per minute, or one gigawatt annual throughput. The excitement of the solar field is that not only are major gains being made, but also that so many of the technologies have the potential of being disruptive game-changers to the energy industry. One year from now, we may be thinking there is enough “pow” in solar power to reach energy parity well before 2015. n

Researchers have big hopes for another relatively new technology that theoretically can reach efficiency levels of more than 44 percent: multi exciton generation. MEG amplifies the effect of a single photon and allows it to generate many electrons, and while the phenomenon mainly has been seen in the lab with quantum dots, rods and wires, Solexant Inc. is attempting to demonstrate that the energy can actually be harvested. The Univ. of Colorado hopes to reach 45 percent efficiency with a separate exciton “fission” approach. Some emerging technologies are aimed at capturing and making more efficient use of solar grade silicon. MIT has a grant to create thin, long lasting silicon wafers without the use of a saw, and Mayaterials Inc. will be working to ramp up a method to derive silicon from agricultural by-products such as rice hulls for less than $25 per kilogram. American Ceramic Society Bulletin, Vol. 87, No. 7

Rochester Institute of Technology will be using a quantum dot technology to expand useable solar spectrum.

33

Low Thermal Expansion Sintered LZSA Glass-Ceramics Lithium-Zirconium-Silicon glass-ceramic systems high coefficient of thermal expansion makes it inadequate for many applications, but the coefficient can be lowered by the addition of alumina.

Glass-ceramics are polycrystalline materials that contain residual glassy phases, obtained by melting glass and forming it into components that are subjected to controlled crystallization.1–4 These materials have several interesting and wide-ranging properties. Glass-ceramics are applied in various areas of society and industry,5 such as ceramic filters. These materials are used in the ceramic-tile field.6 Successful examples include a sintered glass-ceramic product under the Japanese trademark “Neopariés,” and a ceramic tile coated with a glass-ceramic glaze under the Italian trademark “Enduro.” Marazzi Ceramic Industry produces the latter using a process called Fire Stream.

Chemical durability (310–6 mol of Li)

O.R.K. Montedo, F.M. Bertan, R. Piccoli, D. Hotza, A.N. Klein and A.P.N. de Oliveira

Time (min)

The Li2O-ZrO2-SiO2 glass-ceramic system is Fig. 2 Chemical durability of investigated glasses as a function of attack practicularly interesting because the glass obtained time in distilled water, where (●) is LZS1A, (■) is LZS2A, (▲) is LZS4A and and the crystalline phases formed – zircon, (ZrSiO4), (3) is LZS6A. and lithium disilicate, (Li2Si2O5) – provide imporCTE of 11 × 10–6/°C. The result is a new glass-ceramic systant properties for application as ceramic-tile glazes. The authors6–10 have conducted previous research that shows tem named LZSA, Li2O–ZrO2–SiO2–Al2O3 that has a lower CTE. that certain compositions of the LZS system exhibit, in particular, high bending strength as well as high abrasion The objective of this work is to evaluate the influence of and chemical resistance, when compared with traditional substitution of ZrSiO4 by Al2O3 to obtain b-spodumene(ss)materials. based glass-ceramic materials with CTEs between 4.0 × LZS glass-ceramics are naturally white materials that accept coloration by the introduction of an inorganic pigment. They can be produced at 800°C–900°C in 35–60 min using the conditions and machines of a traditional ceramic factory. However, this system is limited because of its high coefficient of thermal expansion, i.e., 9.0 × 10–6/°C to 11.0 × 10–6/°C, compared with that of the ceramic support, i.e., 5.5 × 10–6/°C to 7.0 × 10–6/°C. Therefore, we tested the partial substitution of ZrSiO4 by Al2O3 to form the b-spodumene (solid solution) Li2O∙Al2O3∙(4–10)SiO2 crystalline phase, because the global CTE of the ceramic system depends on the intrinsic properties of each crystalline phase present. The Li2O∙Al2O3∙4SiO2 b-spodumene crystalline phase has a low CTE, 0.9 × 10–6/°C. Appropriate crystallized amounts of b-spodumene compensate the effects caused by the crystalline phases ZrSiO4 with its CTE of 4 × 10–6/°C and Li2Si2O5 with its 34

10–6/°C and 6.0 × 10–6/°C.

The Experiment Tables 1 and 2, respectively, show the formulated and chemical compositions of the investigated LZSA glass-ceramic. We prepared compositions from various amounts of Li2CO3, ZrSiO4, SiO2 and spodumene as raw materials by partial substitution of ZrO2 by Al2O3. We used a glass-ceramic with the composition Table 1 Formulated LZSA Glass-Ceramic Compositions Glass LZS1A LZS2A LZS4A LZS6A

Li2O (wt%) 11.6 11.6 11.7 11.8

SiO2 (wt%)

ZrO2 (wt%)

Al2O3 (wt%)

67.7 68.2 68.6 69.1

20.7 16.8 12.6 8.6

0.00 3.40 7.10 10.50

American Ceramic Society Bulletin, Vol. 87, No. 7

Relative density (%)

Linear shrinkage (%)

Temperature (°C) Fig. 4 Thermal linear shrinkage curve of investigated glass compositions, where (♦) is LZS1A, (■) is LZS2A, (▲) is LZS4A and (×) is LZS6A.

11.6Li2O∙20.7ZrO2∙67.7SiO2 as a reference. We placed ~1 kg of each batch into mullite crucibles and melted them at 1480 ± 3°C for 2 h in a gas furnace. We poured the melts into a ceramic mold to obtain small glass pieces for thermal expansion measurements. We determined the CTEs and the glass transition temperatures using a Model DIL 402C, Netsch, Bayern, Germany, dilatometer under dry air at a heating rate of 20°C/min. We quenched the remaining melts in water and dried them. We then dry-crushed the obtained glass frits in porcelain ball mills for 3 h and sieved them to yield a powder. We determined that the average particle size of the powder was 6 μm using a Model 1064L, Cilas, Orleans, France, laser-scattering particle-size analyzer. We determined chemical composition of the frits using a Model PW2400 X-ray fluorescence spectroscope manufactured by Philips, Eindhoven, Netherlands. We determined the crystallization temperatures of the glass powders using differential thermal analysis under dry air at 20°C/min in a Netzsch Model STA 409 EP hightemperature thermoanalyzer with an empty alumina crucible as reference material. We investigated the amorphous nature of the powdered samples of the as-quenched glasses and the crystalline phases formed during heat-treatments using a Philips Model PW 3710 computer-assisted X-ray powder diffractometer with CuKa radiation. We evaluated chemical resistance according to ISO/R 719-1968. We humidified the glass powders at 9 wt% and then uniaxially pressed them in a hydraulic press at 40 MPa to 1.4 g/cm3 green density in a 12 mm-diameter steel die to study sintering effects. We dried the resulting green samples in a stationary dryer at 120 ± 3°C for 2 h. We isothermally sintered the dried samples in a laboratory electric furnace that could be controlled to ±5°C under air for 60–100 min at 550°C–900°C. We then air-quenched the sintered samples to room temperature. We measured the thermal linear shrinkage of each green and sintered sample at selected temperatures. We measured the dimensions of the green and sintered samples using a screw micrometer. We measured the theoretical density, ρt of the sintered samples using a picnometer and the apparent density, ρap, using the American Ceramic Society Bulletin, Vol. 87, No. 7

Temperature (°C) Fig. 5 Relative density curve of investigated glass compositions, where (♦) is LZS1A, (■) is LZS2A, (▲) is LZS4A and (×) is LZS6A.

Archimedes principle in water immersion at 20°C. We calculated the relative density, ρr, using the apparent and theoretical density measurements. We measured the CTEs of the sintered samples using a Netsch Model DIL 402C dilatometer at a heating rate of 20°C/min. We transversally cut, ground, polished with 1 μm alumina paste and etched in 2% HF for 25 s the sintered samples. Subsequently, we coated the samples with a thin gold film before observing them in a Philips Model XL-30 scanning electron microscope.

Characterization of Glasses Fig. 1 shows XRD patterns of the investigated glass frits show that the obtained glass frits are amorphous, even though we detect low-intensity peaks related to the ZrSiO4, ZrO2 and Al2O3 crystalline phases. These peaks occur because of devitrification or contamination. Such contamination or partial crystallization help the crystallization process,11 because they act as nucleation sites. Budnikov and Pivinskii12 report that 1.5 wt% Al2O3 concentration promotes crystallization. Fig. 2 shows results of chemical durability tests of the investigated glasses. In general, glasses present high chemical durability. However, even the best glass is not completely stable and can be corroded by a given solvent. Chemical durability depends on the temperature, solvent type, contact time between the glass and the solvent, and chemical composition of the glass. The nature and intensity of the chemical bonds that exist in the glass are important. Pannhorst13 reports that the substitution of Si4+ by Al3+ in the glass structure can be achieved over a large range of composition when a compensation of charge is done by at least one of the following ions: Li+, Mg2+ or Zn2+. Therefore, in this system, Al2O3 acts as a network former when substituted for SiO2. However, this does not always occur. Varshneya14 reports that the structural configuration depends on the Al2O3/M2O molar ratio, where M is the alkaline cation, in this case Li1+. When the molar ratio is LZS6A > LZS2A > LZS1A. Fig. 2 confirms this result, because partial substitution of ZrO2 by Al2O3 causes a decrease in the lithium content dissolved in water.

of alkaline metals in water. For instance, low chemical durability can cause a meaningful compositional modification during the wet-grinding process. Table 3 shows that the partial substitution of ZrO2 by Al2O3 changes glass density. Oliveira15 reports that zirconium, because it is an intermediate network former in the glass structure and has high field strength, increases the number of oxygen bonds per silicon atom, which makes the glass structure denser. But, as we discussed earlier, the Al3+ ion acts as a network former in the investigated glasses, but its influence is weaker. In fact, aluminum has a smaller coordination number than zirconium. Therefore, the rupture of chemical bonds between silicon and oxygen atoms results in a more open structure. Besides, there is a significant difference between atomic masses of zirconium (91.22 amu) and oxygen (26.98 amu). Consequently, we expect a decrease of the glass density, as shown in Table 3. Fig. 3 shows that a more open structure, caused by the partial substitution of ZrO2 by Al2O3, also interferes with the thermal properties of the investigated compositions. Table 4 shows that the glass transition temperature decreases with the increase of the substitution of ZrO2 by Al2O3 in the investigated compositions.

Chemical durability is an important property related to glasses, because it represents susceptibility to the solubility

Two factors may have caused the lower Tg.

Table 4 Thermal Properties of Investigated Compositions Glass

Glass transition temperature, Tg (°C)

LZS1A

558

767

884

621

993

209

0.66

LZS2A

554

758

891

605

961

204

0.67

LZS4A

†

540

751

751

933

211

0.68

LZS6A

504

745

745

945

241

0.64

Crystallization Crystallization Softening Melting temperature 1, temperature 2, temperature, temperature, Tc1 (°C) Tc2 (°C) Ts (°C) Tm (°C) Tc – Tg (°C)

Tg/Tm

• The compositional adjustments made in the proposed glass compositions – because of the use of natural raw materials – result in higher alkali contents, including Na2O, K2O and Li2O, as shown in Table 2. These alkalis or oxides act as network modifiers during glass formation, which decreases glass viscosity. Therefore, Tg is achieved at lower temperatures.

• The ZrO2 is partially substituted by Al2O3. Oliveira15 reports a wide composi† tional range in the LZS Estimated by DTA. system and increasing ZrO2 content in the glass composition – keeping the Li2O/SiO2 ratio constant – increase glass viscosity. Zirconium is an intermediate oxide. However, depending on its concentration in the glass composition, it acts as a network former, because it establishes more bonds between oxygen and silicon atoms. Therefore, glasses that contain Zr4+ ions have a denser structure and 2q (deg) a higher Tg. Oliveira Fig. 13 XRD patterns of investigated glass compositions heat-treated at 850°C for 10 min: (a) LZS1A; observes that Tg increases (b) LZS2A; (c) LZS4A; and (d) LZS6A, where E is β-spodumene(ss), D is Li2Si2O5, M is Li2SiO3, Q is from 460°C to 565°C as a 559

α-quartz, T is tridimite, Z is ZrO2 and ZS is ZrSiO4. 36

American Ceramic Society Bulletin, Vol. 87, No. 7

Relative density (%)

density as a function of temperature. Fig. 7 also shows that the maximum densification rate for the investigated glass compositions occurs at 675°C, 670°C, 660°C and 620°C, respectively, for LZS1A, LZS2A, LZS4A and LZS6A compositions. These temperatures are in good agreement with those related to the beginning of crystallization obtained from the DTA curves in Fig. 3. These temperatures show that the densification rate decreases because of the beginning of the crystallization. A progressive increase of the system viscosity must occur at these temperatures. The main mechanism of mass transport during sintering of vitreous materials is viscous flow.16 Therefore, the viscosity increase should inhibit the sintering process and cause a decrease in the densification rate.

Temperature (°C) Fig. 15 Crystallinity of glass samples heat-treated at various temperatures for 10 min, where (♦) is LZS1A, (■) is LZS2A, (▲) is LZS4A and (×) is LZS6A.

result of ZrO2 addition. As we noted earlier, the Al3+ ion also acts as a network former oxide in the investigated compositions, although its action in the glass structure is less effective in relation to the Zr4+ ion. The aluminum atom has a smaller coordination number than does zirconium. Therefore, the partial substitution of Zr4+ ion by the Al3+ ion in the glass structure causes a breaking of chemical bonds between silicon and oxygen atoms, which makes the structure more open. Consequently, we expect a decrease in glass density. Table 3 shows this effect. Moreover, Table 4 shows that the difference between Tc and Tg, i.e., ∆T, increases with lower Tg. This interferes with the sinterability of the investigated glasses.

Sintering Behavior Sintering occurs before crystallization in the thermal treatment process of a traditional glass-ceramic material obtained from a powder. Therefore, sintering should be concluded before crystallization is begun. Therefore, we must know the effects caused by the partial substitution of ZrO2 by Al2O3 in the investigated compositions during sintering.

Therefore, when the Al2O3 content increases in the investigated compositions, Tg also must decrease so that sinterability is improved. However, Shyu and Lee16 report that this effect is limited by crystallization at the surface of the glass particles because a fast increase of viscosity takes place, which inhibits the sintering process by viscous flow. Therefore, besides the viscosity, the capability of crystallization of the parent glass powder also determines the densification of several glass-ceramic systems.

Crystallization Behavior Rabinovich reports that sintering is improved when crystallization is superficial, because, in volumetric crystallization, sintering occurs at higher viscosities. 17

We evaluate the predominant mechanism of crystallization in the investigated system by comparing two DTA thermograms of the same composition: one obtained from analysis conducted on a monolithic glass sample obtained from the melted composition, and another obtained from glass frit powder. Fig. 8 shows DTA thermograms of monolithic glass samples. Fig. 8 shows the absence of crystallization peaks. This means that the crystallization mechanism for the investigated glass compositions cannot be a volumetric mechanism. On the other hand, if the glass shows a tendency to crystallization, and in this case it is superficial, crystalline phases formed in the external surface of the monolithic glass sample and in the surface of the impurities and pores are in too small a quantity to be detected using DTA.

Figs. 4 and 5 show, respectively, the behavior of the investigated compositions during sintering through thermal linear shrinkage and relative density measurements. Figs. 4 and 5 show that densification starts at lower temperatures as the substitution of ZrO2 by Al2O3 increases. LZS1A densification starts at 640°C, and LZS6A densification starts at ~560°C. However, densification, in practice, is interrupted – actuHowever, Fig. 3 shows that glass powders with a mean ally, the rate of densification decreases almost to zero – at particle size of 5 μm present at least one crystallization peak. ~700°C for all investigated compositions, except for the LZS6A composition, whose interruption occurs Table 5 Coefficient of Thermal Expansion of Investigated Compositions at 660°C. Fig. 6 shows that these temperatures take CTE (×10–6/°C) (25°C to 325°C) into account the behavior of relative density as a ———————————————————————————————————————————————— function of temperature and of its second derivate Glass, Glass, with relation to temperature. Fig. 6 shows that it is measured calculated† 750°C 800°C 850°C possible to obtain with good accuracy the temperaLZS1A 7.98 ± 2.1 7.05 7.68 ± 3.2 9.71 ± 5.1 7.49 ± 2.5 tures of beginning (maximum point) and interruption LZS2A 7.95 ± 3.3 6.84 7.24 ± 4.5 8.01 ± 0.5 7.24 ± 3.7 (minimum point) of densification. Therefore, densification occurs at ~90°C. Fig. 7 shows that we can obtain maximum densification rate from the first derivative curve of the relative American Ceramic Society Bulletin, Vol. 87, No. 7

LZS4A

7.13

6.25 ± 1.9 6.00 ± 1.7

5.62 ± 0.4

LZS6A

7.44

4.60 ± 2.0 4.62 ± 0.8

4.90 ± 0.3

†

8.05 ± 4.6

According to Appen in Ref. 2.

37

Low Thermal Expansion Sintered LZSA Glass-Ceramics These powders have high specific surface area and, at the heating rate used, produce sufficient quantities of crystalline phases to be detected. This means that the predominant crystallization mechanism in the LZSA system glass-ceramic is superficial. Nevertheless, we can determine the mechanism of glass crystallization in other ways. Zanotto18 and James and Jones19 report that the mechanism of crystallization in glasses can be evaluated using the Tg/Tm ratio. They report that the crystallization is volumetric when the Tg/Tm ratio is 0.58. Tg/Tm values in Table 4 show that all investigated compositions crystallize by superficial mechanism. Table 4 also shows that LZS4A composition presents a greater tendency to superficial crystallization with respect to the other investigated compositions. The LZS glass-ceramic system studied by Oliveira15 presents the following crystalline phases: Li2Si2O5, ZrSiO4, SiO2 as α-quartz and lithium metasilicate, i.e., Li2SiO3. In this work, the LZS glass-ceramic composition is modified so that β-spodumene(ss) glass-ceramic materials with low CTE can be obtained. This last mentioned crystalline phase, also named β-keatite(ss), has the general formula Li2O∙Al2O3∙(4–10)SiO2 and presents CTE values

positive and near zero. Mineralogical β-spodumene, i.e., Li2O∙Al2O3∙4SiO2, has a CTE of 0.9 × 10–6/°C between 20°C and 1000°C.20 Figs. 9 to 12 show, respectively, XRD patterns of LZS1A, LZS2A, LZS4A and LZS6A glass-ceramics heat-treated for 10 min at 800°C, 850°C and 900°C. These figures show the main crystalline phases in the investigated glasses: • β-spodumene(ss) Li0–6Al0–6Si2–4O6, per Powder Diffraction File No. 21-503, International Centre for Diffraction Data, Newtown Square, Pa.; • LiAlSi3O8, per File No. 15-27; • ZrSiO4, per File No. 6-266; • Li2Si2O5, per File No. 24-651; • ZrO2, per No. 13-307; • Li2SiO3, per File No. 29-828; • α-quartz, per File No. 5-490; • cristobalite, per File No. 11-695; and • tridimite, per File Nos. 14-260 and 18-1170. The relative heights of the peaks in Figs. 9 to 12 show that, as substitution of ZrO2 by Al2O3 in the parent glass composition increases, the ZrSiO4 content formed decreases and the β-spodumene(ss) content formed increases. ZrSiO4

Fig. 18 SEM photographs of LZS4A glass compositions heat-treated for 10 min at (a) 600°C, (b) 650°C, (c) 700°C and (d) 900ºC after they were etched in 2% HF for 25 s. 38

American Ceramic Society Bulletin, Vol. 87, No. 7

is detected at 850°C in the LZS1A composition, and it becomes the main crystalline phase at 900°C. However, in the other glass compositions, β-spodumene(ss) is the main formed crystalline phase and is detected at 700°C. The Li2Si2O5 crystalline phase is detected at 800°C. However, at 900°C, its participation in the structure of the compositions is decreased strongly and is not detected for LZS4A and LZS6A compositions at 900°C. The SiO2 participation decreases when the temperature is increased. The SiO2, in any identified crystalline phases – α-quartz, cristobalite or tridimite – probably is consumed by β-spodumene(ss) formation. Fig. 13 shows XRD patterns of the effect of partial substitution of ZrO2 by Al2O3 in the structure of the investigated glass compositions heat-treated at 850°C for 10 min. Fig. 13 shows that, as the Al2O3 content in the investigated glass compositions increases, the Li2SiO3 and β-spodumene(ss) crystalline fractions increase and the ZrSiO4, Li2Si2O5, α-quartz and tridimite crystalline fractions formed decrease. The formation of β-spodumene(ss) and α-quartz crystalline phases at 700°C. Indicates that densification is interrupted at this temperature. The beginning of crystallization at 700°C interferes meaningfully in the thermal linear shrinkage rate. Therefore, the obtained materials exhibit lower relative density values. As we discussed earlier, crystallization in this system occurs superficially, because the glass powders have high specific surface area. Therefore, we can state that the high crystallization rate prevents densification progress. Fig. 14 shows the beginning of the crystallization process at 700°C for LZS1A glass. The same behavior is observed in the other glass compositions. Fig. 15 shows the crystallinity of the investigated glass compositions at several temperatures determined using the Ohlberg method.21 Fig. 15 shows that the crystallinity of glass compositions heat-treated for 10 min increases as the temperature and Al2O3 content are increased, except for the LZS6A glass composition, which shows, at 900°C, lower crystallinity with respect to the other glass compositions. We evaluate the kinetics of crystallization for the LZS4A glass composition at 950°C from the isotherms shown in Fig. 16. Analysis of the relative peak heights of the β-spodumene(ss) and ZrSiO4 crystalline phases shows that there is no significant variation in the amounts of these formed phases in the time range studied. In other words, the kinetics of crystallization strongly depends on the temperature and is almost invariable with respect to the heattreatment time. This behavior is related to the high specific surface area of the glass powders. Fig. 17 shows that the crystallinity determined using the Ohlberg method21 is almost constant in the heat-treatment time range.

Microstructural Analysis As we discussed earlier, the thermal linear shrinkage of the LZS4A glass composition occurs at 600°C – 700°C, and the maximum densification rate occurs at 660°C. Fig. 18, which shows the progress of the microstructure as the sinterAmerican Ceramic Society Bulletin, Vol. 87, No. 7

Fig. 19 SEM photograph of LZS4A glass composition heattreated at 900ºC for 10 min after it was etched in 2% HF for 25 s.

ing temperature increases for the LZS4A glass composition, indicates good agreement with sintering discussions regarding this glass-ceramic system. Fig. 18 shows that powder particles maintain their identities at 650°C. However, at 700°C, the particles form a monolithic body. Fig. 18(d) shows that, at 900°C, the glassceramic microstructure has less porosity and a small crystalline phase in the form of light-colored particles that are well distributed throughout the structure. Fig. 19 shows the LZS4A glass composition with crystalline phases in the form of light-colored particles that are distributed throughout the microstructure. The particles vary in size from 1 µm to 3 µm. Fig. 20 shows, using microchemical analysis with an EDS detector, that the chemical elements present are silicon, aluminum and oxygen. Lithium could not be detected using this technique because of its low atomic number. However, the crystalline phase observed in Fig. 19 probably is a lithium aluminum silicate, β-spodumene, also identified in the XRD patterns discussed early. Fig. 21 shows SEM photographs of the LZS2A, LZS4A and LZS6A glass compositions sintered at 850°C for 10 min. Fig. 21 shows that porosity decreases because of substitution of ZrO2 by Al2O3. This behavior is in good agreement with the relative density results and can be associated to chemical composition changes by partial alumina dissolution in the glasses.

Coefficient of Thermal Expansion Glass-ceramic materials can be used in many applications. For example, they can be used as a component of a doublelayered material, even if stress must be avoided in the layer interfaces. In these cases, the CTE of the materials, among others features, needs to be evaluated. Therefore, we must often design compositions in which the crystalline phases intrinsically show low CTE. The main crystalline phases formed in the LZSA system are β-spodumene(ss) and ZrSiO4. We expect that, as the 39

Low Thermal Expansion Sintered LZSA Glass-Ceramics β-spodumene(ss) fraction increases in the glass matrix, the CTE of the resulting glass-ceramic material decreases. Table 5 shows that CTE decreases with the partial substitution of ZrO2 by Al2O3 in the investigated glass compositions. However, Figs. 9 and 10 show that the CTEs of LZS1A and LZS2A glass compositions increase at 800°C because of formation of the Li2Si2O5 crystalline phase that has a high CTE,22 11.0 × 10–6/°C.

Glass-Ceramics Produced, Studied We produce glass-ceramics based on the LZSA system by sintering and crystallizing a glass powder. Sintering starts at ~570°C and is completed at 700°C. However, we obtain a 93% relative density and an 18% percent thermal linear shrinkage at lower temperatures by increasing the partial substitution of ZrO2 by Al2O3. On the other hand, we obtain a maximum relative density and thermal linear shrinkage of LZS4A glass composition at 650°C. This allows us to produce a material at lower temperatures. In this case, crystallization starts at 650°C, and β-spodumene(ss) is the main identified crystalline phase. Therefore, the CTE drastically decreases by an increase in the partial substitution of ZrO2 by Al2O3. The LZS6A glassceramic composition has the most interesting CTE, i.e., 4.6 × 10–6/C at 850°C. n

Z. Strnad, Glass Science and Technology; pp 9, 85, 114, 166. Elsevier, New York, 1996. 4

J. M. Rincon and M. Romero, Mater. Constr., 46 [242–243] 91 (1996). 5

A.P. Novaes de Oliveira, et al., J. Am. Ceram. Soc., 79 [4] 1092 (1996). 6

7 A.P. Novaes de Oliveira, et al., Thermochim. Acta, 286, 375 (1996). 8 A.P. Novaes de Oliveira, et al, Phys. Chem. Glasses, 39 [4] 213 (1998).

A.P. Novaes de Oliveira, et al, Phys. Chem. Glasses, 41 [2] 100 (2000). 9

A.P. Novaes de Oliveira, et al., J. Mater. Sci., 36, 1 (2001). 10

E.M. Rabinovich, J. Mater. Sci., [391] 4259–97 (1985).

11

P.P. Budnikov and Y.E. Pivinskii, Russ. Chem. Rev., 36, 210 (1967). 12

W. Pannhorst, “Overview”; Ch. 1, pp 1–12 in BACH H, Low Thermal Expansion Glass Ceramics. Springer, Germany, 1995. 13

A.K. Varshneya, Fundamentals of Inorganic Glasses. Academic Press, New York, 1994. 14

A.P. Novaes de Oliveira, “Progettazione, Caratterizzazione ed Ottenimento di Vetri-Vetroceramici Appartenenti al Sistema Li2O–ZrO2–SiO2”; Ph.D. Thesis. Douttorato di Ricerca in Ingegneria dell’Informazione e dei Materiali, Universitá Degli Studi di Modena, Modena, Italy, 1997. 15

Editor’s Note Additional Figures and Tables related to this article are available online at ceramicbulletin.org.

Acknowledgments The authors are grateful to Capes and CNPq/Brazil for funding this work.

About the Authors O.R.K. Montedo, F.M. Bertan and R. Piccoli are research staff members with SENAI/CTCmat – Center of Technology in Materials, Criciúma, S.C., Brazil. D. Hotza is a faculty member with the Dept. of Chemical Engineering, Federal University of Santa Catarina, Florianópolis, S.C., Brazil. A.N. Klein and A.P.N. de Oliveira are faculty members with the Dept. of Mechanical Engineering, Federal University of Santa Catarina, Florianópolis, S.C., Brazil.

References E.D. Zanotto; p 65 in Nucleation and Crystallization in Glasses and Liquids. American Ceramic Society, Westerville, Ohio, 1993. 1

J.J. Shyu and H.H. Lee, J. Am. Ceram. Soc., 78 [8] 2161–67 (1995). 16

17 E.M. Rabimovich, “Cordierite Glass-Ceramics Produced by Sintering”; pp 327–33 in Advances in Ceramics, Vol. 4, Nucleation and Crystallization in Glasses. American Ceramic Society, Columbus, Ohio, 1982.

E.D. Zanotto, J. Non-Cryst. Solids, 89, 361–70 (1987).

18

P.F. James and R.W. Jones, Glass Ceramics; pp 102–13. Edited by M. Cable and J.M. Parker. High-Performance Glasses, New York, 1982. 19

M. Haigh, Am. Ceram. Soc. Bull., 76 [4] 75–78 (1997).

20

R. Roesky and J.R. Vagner, J. Am. Ceram. Soc., 74 [5] 1129–30 (1991). 21

Z. Strnad, Glass-Ceramic Materials. Elsevier, New York, 1986. 22

J.M.F. Navarro, El Vidrio; pp 51–52. CSIC, Madrid, Spain, 1991. 2

3 A.K. Varshneya, Fundamentals of Inorganic Glasses; pp 13, 82. Academic Press, New York, 1994.

40

American Ceramic Society Bulletin, Vol. 87, No. 7

2q (deg) Fig. 1 XRD patterns of investigated glass frits: (a) LZS1A; (b) LZS2A; (c) LZS4A; and (d) LZS6A, where A is Al2O3; ZS is ZrSiO4 and Z is ZrO2.

Temperature (°C)

Fig. 3 DTA curves of investigated compositions: (a) LZS1A; (b) LZS2A; (c) LZS4A; and (d) LZS6A.

d2Drel/dT2

Temperature (°C)

dDrel/dT

Fig. 6 Second derivative curve of relative density as a function of temperature, where (♦) is LZS1A, (■) is LZS2A, (▲) is LZS4A and (×) is LZS6A.

Temperature (°C)

∆T

Fig. 7 First derivative curve of the relative density as a function of the temperature, where (♦) is LZS1A, (■) is LZS2A, (▲) is LZS4A and (×) is LZS6A.

Temperature (°C) Fig. 8 Monolithic glass sample thermograms of investigated glass compositions: (a) LZS1A; (b) LZS2A; (c) LZS4A; and (d) LZS6A.

2q (deg) Fig. 9 XRD patterns of LZS1A glass composition heat-treated for 10 min at (a) 800°C, (b) 850°C and (c) 900°C, where E is β-spodumene(ss), D is Li2Si2O5, M is Li2SiO3, T is tridimite, Z is ZrO2 and ZS is ZrSiO4.

2q (deg) Fig. 10 XRD patterns of LZS2A glass composition heat-treated for 10 min at (a) 800°C, (b) 850°C and (c) 900°C, where E is β-spodumene(ss), D is Li2Si2O5, M is Li2SiO3, Q is α-quartz, T is tridimite, Z is ZrO2 and ZS is ZrSiO4.

2q (deg) Fig. 11 XRD patterns of LZS4A glass composition heat-treated for 10 min at (a) 800°C, (b) 850°C and (c) 900°C, where E is β-spodumene(ss), D is Li2Si2O5, M is Li2SiO3, C is cristobalite, Z is ZrO2 and ZS is ZrSiO4.

2q (deg) Fig. 12 XRD patterns of LZS6A glass composition heat-treated for 10 min at (a) 800°C, (b) 850°C and (c) 900°C, where E is β-spodumene(ss), M is Li2SiO3, C is cristobalite, Z is ZrO2 and ZS is ZrSiO4.

2q (deg) Fig. 14 XRD patterns of LZS1A glass composition heat-treated for 10 min: (a) amorphous; (b) 650°C; (c) 700°C; and (d) 750°C, where E is β-spodumene(ss), Q is α-quartz, Z is ZrO2 and A is Al2O3.

2q (deg) Fig. 16 XRD patterns of LZS4A glass composition heat-treated at 950°C for (a) 0 min, (b) 5 min, (c) 15 min and (d) 60 min, where E is β-spodumene(ss), Q is α-quartz, ZS is ZrSiO4 and M is Li2SiO3.

Crystallinity (%)

Time (min) Fig. 17 Crystallinity of LZS4A glass composition heat-treated at 950°C for various time intervals.

Fig. 20 Partial chemical composition obtained by EDS detector for the crystalline phase identified in Fig. 19.

Fig. 21 SEM photographs of selected investigated compositions heat-treated at 850°C for 10 min: (a) LZS2A; (b) LZS4A; and (c) LZS6A, after they were etched in 2% HF for 25 s.

Table 2 Chemical Composition of Obtained LZSA Glass-Ceramics Glass

Oxide (wt%) ——————————————————————————————————————————————————————————————————————————————————— SiO2 Al2O3 ZrO2 Li2O K2O Na2O TiO2 Fe2O3 CaO MgO P 2O 5

SiO2/ Li2O†

ZrO2/ Al2O3†

LZS1A

64.07

7.89

18.21

9.12

0.07

0.04

0.10

0.05

0.23

0.13

0.07

78/22

2.31

LZS2A

62.12

12.57

15.50

8.57

0.16

0.15

0.09

0.08

0.29

0.20

0.22

78/22

1.23

LZS4A

61.61

15.29

12.23

8.84

0.28

0.36

0.08

0.12

0.46

0.22

0.48

78/22

0.80

LZS6A

63.74

15.33

9.16

9.22

0.29

0.46

0.07

0.43

0.45

0.17

0.64

77/23

0.60

†

Molar basis.

Table 3 Measured and Calculated Densities of Investigated Compositions† Glass

ρmeasured (g/cm3)

ρcalculated (g/cm3)

ρcalculated (g/cm3) 2.64

LZS1A

2.64 ± 1.7

2.66

LZS

2.68

2.68

LZS2A

2.57 ± 1.5

2.61

2.60

LZS4A

2.52 ± 1.3

2.55

2.53

LZS6A

2.49 ± 1.3

2.49

2.48

‡

According to Appen and Higgins and Sun in Ref. 2. ‡LZS (V-785) glass investigated by Oliveira15 and taken as reference in this work. †