with common natural silica-containing materials, like ball clay, kaolin, and/or diatomite were pre- pared in order to achieve interesting final fired compositions.
Journal of Materials Synthesis and Processing, Vol. 10, No. 6, November 2002 (© 2003)
Mullite-Alumina Refractory Ceramics Obtained from Mixtures of Natural Common Materials and Recycled Al-Rich Anodizing Sludge D. U. Tulyaganov,1 S. M. H. Olhero,1 M. J. Ribeiro,2 J. M. F. Ferreira,1 and J. A. Labrincha1,3
This work describes attempts to reuse Al-rich sludge produced from industrial anodizing and surface treatment processes in the fabrication of mullite-based refractory materials. The complete characterization of the residue is reported, including physical and compositional parameters. Mixtures with common natural silica-containing materials, like ball clay, kaolin, and/or diatomite were prepared in order to achieve interesting final fired compositions. Rheological characterization of the suspensions, their slip casting performance, the thermal behavior of the consolidated bodies, and crystalline phase evolution are also detailed. From the knowledge acquired, pretreatment operations have been suggested to facilitate the recycling of the Al-rich sludge and to improve the characteristics of mullite-based materials. KEY WORDS: Mullite; alumina; Al-anodizing sludge; slip casting.
1. INTRODUCTION
posal. In EC countries, about 100,000 metric tons per year are currently generated [3], and no interesting applications for this sludge have been reported so far. On the other hand, predictable high alumina contents of such calcined residues makes them very attractive for recycling processes, such as the recovery of aluminium-based compounds, and their incorporation in other products (e.g., ceramic pastes), or the formation of alumina-based bodies or powders [4]. In this last field, several materials have been investigated as inertization matrixes, such as concrete, glass, and ceramics [5–8]. The current work describes the potential for producing mullite-alumina refractory materials, by combining the Al-rich sludge with several natural silica-based materials. Ceramic bodies belonging to the mullite-alumina system are widely used in electronic, optical, and hightemperature structural applications [9–11], because of the chemical inertness and mechanical properties of the material. From a theoretical point of view, the study of the mullite region of alumina-silica-alkaline (or alkaline earth) oxide ternary equilibrium phase diagram might give helpful predictions about the dominant phase relationships that may occur after a proper firing step [12].
The aluminium industry is affected by pressure from society to reduce its impact on the environment. This has resulted in tremendous efforts to cut energy consumption and reduce wastes by improving production methods and introducing new technologies. Properly treated and neutralized sludge from aluminium anodizing or powder coating processes are basically constituted by water (70–80%) and colloidal aluminium hydroxides and variable minor elements [1]. Despite their high degree of metal mobility in aqueous solutions, these residues are often classified as nonhazardous [2]. However, the high daily production amounts and the difficulties on reducing their volume by suitable filterpressing methods impose high transportation costs for dis-
1
Ceramics and Glass Engineering Department, CICECO, University of Aveiro, 3810-193 Aveiro, Portugal. 2 ESTG, Polytechnique Institute of Viana do Castelo, 4900 Viana do Castelo, Portugal. 3 To whom all correspondence should be addressed. email: JAL@ CV.UA.PT
311 1064-7562/02/1100-0311/0 © 2003 Plenum Publishing Corporation
312
Tulyaganov, Olhero, Ribeiro, Ferreira, and Labrincha
However, the nature of the starting materials and/or the use of different processing routes significantly affect the type of chemical reactions occurring during sintering, the crystalline phase formed, and the ultimate material properties [13]. To obtain properly sintered materials, it is very important to know the nature and properties of starting components and ensure perfect control of all the relevant parameters of the processing steps. In the present work mullite-alumina ceramic materials have been prepared by slip casting from aqueous suspensions of alumina-rich sludge, kaolin, and ball clay. Diatomaceous earth, which is normally extracted as a byproduct of sand mining, was also used as a cheap alternative silica-source material.
2. EXPERIMENTAL 2.1. Raw Materials and Batch Formulations The main component used was an Al-rich sludge generated either from aluminium anodizing or powder coating industrial processes (Extrusal, S. A., Aveiro, Portugal). The Al-rich sludge was amorphous and mostly composed of aluminium hydroxide and sulphide [1]. Other sulphides (Ca and Na) were also present as minor constituents. Its moisture content might reach 85 wt.%. Drying (110°C for 24 h) and calcining (1350°C for 2 h) steps were conducted to remove the water and partially decompose some soluble salts, which would otherwise cause detrimental effects during processing (coagulation of the suspensions, segregation effects during drying, high shrinkage values during sintering) and to the final material properties. XRD analysis revealed the formation of a-Al2O3 above 1100°C. The physical and toxicological characterization, as well as the thermal behavior of the received material was reported elsewhere [1, 8]. The knowledge acquired is essential to define the beneficial pretreatment operations that facilitate its reuse. The mean particle size of the calcined sludge after ballmilling for 15 h was 2.22 m.
Three different natural raw materials (alone or combined with each other) were used as silica sources: (i) diatomite (Sociedade Anglo-Portuguesa de Diatomite, Óbidos, Portugal); (ii) ball clay “gray-T2” (ADM, Pombal, Portugal); and (iii) high-grade kaolin “Standard Porcelain (SP)” (ECC, UK). The diatomite has been previously calcined (600°C for 2 h) to eliminate significant amounts of organic matter present in the raw material. It is mainly composed of amorphous silica, but the relatively high content of iron oxide strongly limits its use in the production of whiteware common ceramics. The average mineralogical composition (wt.%) of kaolin includes kaolinite (84%), micaceous phases (13%), some feldspar (1%), and other minerals (2%). In comparison, ball clay contains higher amounts of micaceous minerals and free quartz. The average particle sizes (Coulter LS 230, UK) of the material was: 11.0 m (diatomite), 5.3 m (ball clay), and 6.1 m (kaolin). Chemical data determined by XRF (Philips X’Unique II, Netherlands) or as given by the supplier are presented in Table I. The starting batch formulations are shown in Table II, and the corresponding oxide compositions estimated from the data given in Table I are presented in Table III. 2.2. Experimental Techniques Suspensions containing different solids’ weight fractions were prepared in the presence of different amounts of a commercial dispersant (Dolapix CE 64, Zshimmerand Schwarz, Germany). Stress sweep measurements were performed with a rotational stress controlled rheometer (Carri-Med 500 CSL, UK) using a cone and plate-measuring configuration. Based on the rheological characterization, the optimal amount of deflocculant and solids loading could be selected to prepare the suspensions for slip casting. The batch components were added to the dispersant solution under stirring, according to the following order: kaolin, ball clay, Al-sludge, and diatomite (when used). Deagglomeration and milling was subsequently conducted for 1 h in a rapid ball mill using alumina grinding media. A deairing step that consisted of
Table I. Chemical Composition of Raw Materials, Determined by XRF or as Given by the Supplier Raw material Al-sludgea Diatomiteb Ball-clayc Kaolin SPc a
P2O5
SiO2
TiO2
Al2O3
Fe2O3
Cr2O3
CaO
MgO
MnO
K2O
Na2O
LOI
— — 0.04 —
2.38 87.20 56.54 48.00
0.05 — 1.08 0.02
91.39 2.60 30.37 37.00
1.13 3.40 1.66 0.65
0.42 — — —
2.44 1.70 0.20 0.07
0.32 — 0.98 0.30
0.02 — 0.006 —
0.02 5.10 1.58 1.60
1.92 — 1.08 0.10
— — 6.42 12.50
Calcined at 1400°C for 2 h. Calcined at 600°C for 2 h. c As given by the supplier. b
Mullite-Alumina Refractory Ceramics Obtained from Mixtures Table II. Tested Batch Formulations (wt.%) Formulation Al-sludge Diatomite Ball clay Kaolin SP
A
B
C
60.6 7.4 6.5 25.5
64.2 — 7.2 28.6
70.0 — 6.0 24.0
rolling the slips in polyethylene containers (without balls) in a slow rotating system for 12 h was followed. Slip cast samples were consolidated by pouring the suspensions inside plastic rings placed on a plaster plate, and then successive drying at two temperature stages of 40° and 100°C. The dried samples were sintered at three different temperatures (1300°, 1400° and 1500°C) by keeping 1.5 h at the maximum value and using 5°C/min as heating and cooling rates. Apparent densities of green and fired bodies were determined by the Archimedes method (Hg immersion). Diametral shrinkage upon drying and firing, water absorption, and 4-point bending strength (Autograph AG-A, Shimadzu, Japan) of the sintered samples, were determined. For the bending test, uniaxially pressed bars (4 4 50 mm) were prepared. Dilatometry (Netzsch 402EP, Germany) was used to characterize the thermal expansion behavior of fired samples. Mineralogical analysis of the raw materials and crystalline phases formed during firing was performed by XRD (Rigaku Denk Co., Japan). Microstructures of the sintered bodies were observed by SEM (Hitachi, Tokyo, Japan) and their chemical composition estimated by coupled EDS system (S60 DX90). 3. RESULTS AND DISCUSSION 3.1. Suspension Preparation and Characterization
313
studied for suspensions of formulation A containing 40 wt.% solids. Figure 1 shows that all suspensions exhibit a plastic behavior with defined yield stress values. It can be seen that the shear stress at any given shear rate decreases as the amount of deflocculant increases up to 0.4 wt.%, increasing again with further additions. This indicates that the maximum dispersing effect at this solids loading seems to occur for 0.4 wt.% of Dolapix. Based on these results, this concentration of dispersant was selected as the lower limit to prepare more concentrated (60 wt.% solids) slurries. Because the optimal amount of dispersant depends on the solids loading [14], further refinements of the effect of the amount of dispersing agent on flow ability of suspensions were also studied at 60 wt.% solids. Figure 2 shows that the best dispersion effect has been achieved for 0.5 wt.% Dolapix. However, even under these conditions, the tixotropic area (the area between the up and down flow curves) is still noticeable corresponding to an observed trend for suspension structuration. Although these characteristics are commonly found in clay-based suspensions, in the present case they should derive mainly from two factors: (i) soluble salts of the Al-sludge (the major component), which will reduce the range of repulsive interaction forces between particles [15]; and (ii) the presence of particles with different surface chemistry (the isoelectric point of silica is around pH 2, while that of alumina is around pH 9) [16]. Therefore the comparison of the rheological behavior of different formulations was conducted for suspensions containing 55 wt.% solids and dispersed with 0.5 wt.% Dolapix. The results shown in Fig. 3 confirm the previously observed plastic behavior, with formulation B approaching more the Bingham model with a higher and well-defined yield stress. This might be attributed to its slightly higher amount of ball clay. Although the amount of the containing soluble salts com-
The effect of the amount of dispersant on the flow behavior (shear stress vs. shear rate curves) was first Table III. Main Oxide Constituents (wt.%) of Tested Batches Estimated from the Chemical Data Presented in Table I Formulation SiO2 Al2O3 Fe2O3 TiO2 CaO MgO K2O Na2O Cr2O3 MnO
A
B
C
24.70 69.48 1.26 0.10 1.70 0.34 0.92 1.30 0.27
20.11 74.45 1.08 0.12 1.66 0.38 0.60 1.39 0.29
17.16 77.29 1.09 0.09 1.80 0.36 0.50 1.47 0.32
Fig. 1. The effect of the amount of dispersant on the flow behavior of suspensions of formulation A containing 40 wt.% solids.
314
Fig. 2. The effect of the amount of dispersant on the flow behavior and tixotropic area of suspensions of formulation A containing 60 wt.% solids.
ponent increases from formulation A to C, the fluidity of the slurries is improved for shear rates higher than about 50 s1, (B), and along the whole shear rate range (C). This observation supports the hypothesis about the heteroflocculating effect among the suspended particles having different surface chemistry characteristics in the case of formulation A, indicating that the absence of diatomite enhances fluidity.
3.2. Effects of the Dispersion Conditions on Slip Casting Performance and on the Properties of the Sintered Bodies Table IV and Fig. 4 report the effects of the amount of dispersant and solids loading on the characteristics of green and sintered bodies prepared by slip casting from formulation A. For a question of clarity, only the most significant results about the evolution of apparent density and water absorption as a function of sintering temperature have been displayed in Fig. 4. It can be seen that at
Tulyaganov, Olhero, Ribeiro, Ferreira, and Labrincha 40 wt.% solids, the better slip casting performance seems to have been achieved for 0.2 wt.% Dolapix, corresponding to underdeffloculating conditions according to Fig. 1. At this low solids loading, segregation phenomena are likely to occur that will lead to concentration gradients and overall less dense compacts during slip casting and should be enhanced by increasing the fluidity of the slurries [17, 18]. This explains why the values of green and sintered properties were not improved beyond 0.1– 0.2 wt.% dispersant. At 60 vol.% solids, particle segregation is hindered and the slip casting performance was significantly improved. The higher degree of homogeneity achieved at this solids loading enabled higher sintered densities and lower values of water absorption. Further, the best results could be obtained from the highly dispersed (0.5 wt.% Dolapix) suspension, according to the rheological results (Fig. 2). Figure 4 shows that although water absorption continuously decreases along the whole temperature range, apparent density shows a decrease between 1300° and 1400°C, increasing afterward. The decrease between 1300° and 1400°C can be attributed to the formation of lower-density crystalline phases such as mullite, detected by XRD. With solids loading increasing to 60 wt.%, viscosity increases and the extent of particles segregation decreases, permitting more homogeneous and denser green and sintered bodies. The highest density and lowest water absorption values were obtained in the presence of 0.5 wt.% Dolapix in close agreement with flow curves shown in Fig. 2. As for the rheological behavior (Fig. 3), comparison of slip casting performance between different formulations was made using suspensions with 55 wt.% solids dispersed with 0.5 wt.% Dolapix (Table V). It can be seen that the values of green and sintered density are significantly higher for formulations B and C, that is, in the absence of diatomite. These results are also in good agreement with the rheological measurements, pointing again to a heterocoagulation effect, and indicate that the structure of particles in suspension is transmitted to the wall being deposited by slip casting. The gradual enrichment in alumina from compositions A to C may also contributed to the enhanced density values of the samples B and C.
3.3. Influence of Composition and Sintering Temperature on Crystalline Phase Formation Fig. 3. Rheological behavior of suspensions of the formulations A, B, and C containing 55 wt.% solids and 0.5 wt.% Dolapix CE 64.
Although the overall compositions of formulations A, B, and C are very complex, as shown in Table III, a simplification attempt to represent them in the ternary
Mullite-Alumina Refractory Ceramics Obtained from Mixtures
315
Table IV. Effects of the Amount of Dispersant and Solids Loading on Green and Sintered (at Different Temperatures) Characteristics of Bodies Prepared from Formulation A 40 wt.% solids Properties Apparent density (g/cm3) Fired Shrinkage (%) Drying Firing Water absorption (%)
Green 1300°C 1400°C 1500°C Drying 1300°C 1400°C 1500°C 1300°C 1400°C 1500°C
60 wt.% solids
0.1% D
0.2% D
0.4% D
0.7% D
0.4% D
0.5% D
1.48 2.43 2.34 2.52
1.51 2.38 2.32 2.54
1.51 2.38 2.30 2.52
1.50 2.33 2.30 2.50
1.58 2.52 2.44 2.60
1.61 2.58 2.45 2.62
3.4 17.9 17.9 19.0 6.28 4.54 1.55
equilibrium phase diagram of the main components SiO2-Al2O3-CaO [19] was made. Following this approach, similar oxides were grouped and converted into the main representative oxide of each family (e.g., alkaline and earth-alkaline oxides were accounted as CaO, TiO2, as SiO2, etc.). The simplified compositions are then (wt.%): (A) 24.80 SiO2, 71.20 Al2O3, 4.00 CaO; (B) 20.16 SiO2, 75.97 Al2O3, 3.87 CaO; (C) 17.21 SiO2, 78.81 Al2O3, 3.98 CaO. Their locations in the phase diagram are shown in Fig. 5. All of them belong to the alumina-mullite-anorthite triangle and are located in the field of corundum crystallization. Formulation A is closer to the mullite phase, and some alumina enrichment was tried with compositions B and C. XRD results of samples fired at 1500°C (Fig. 6) confirm these tendencies. The predicted higher amount of liquid phase and the lower relative amount of alumina in formulation A were
4.1 18.3 18.3 19.7 6.44 4.57 0.69
4.1 17.4 17.4 18.6 5.90 4.85 0.81
5.4 17.7 17.7 19.2 7.01 4.95 2.04
5.5 19.2 19.2 20.4 4.48 3.17 0.08
5.4 18.4 17.4 18.8 3.28 2.63 0.00
experimentally supported by its stronger tendency for pyroplastic deformation. The presence of very reactive amorphous silica (diatomite) in samples A also leads to a more abundant liquid phase that will hinder the achievement of high density values, even if the relative amounts of fluxing agents such as the alkaline and alkaline-earth oxides, mainly incorporated by the Al-sludge, increase in samples B and C. On the other hand, diatomite particles have a typical intrinsic microporosity that is difficult to eliminate even after firing at very high temperatures [20]. Accordingly, the richer-alumina samples C, exhibit higher densification rates, in agreement with other reports [21, 22]. The increasing maturation degree associated with higher shrinkage and lower water absorption values is responsible for the improved mechanical strength observed with samples B and C. A detailed study of phases evolution upon firing was conducted for formulation A. The XRD spectra (Fig. 7) show that mullite and alumina are the main crystalline phases present at 1300°C. By further heating, the Table V. Characteristics of the Green and Sintered (1500°C for 1.5 h) Bodies Prepared from Suspensions Containing 55 wt.% Solids Properties
Fig. 4. Dependence of apparent density and water absorption on sintering temperature for the slip cast samples of formulation A prepared from the suspension containing 40 and 60 wt.% solids dispersed with different amounts of Dolapix CE 64.
Apparent density (g/cm3) Green Fired Shrinkage (%) Drying Drying Firing Water absorption (%) Bending strength (MPa) Thermal expansion coefficient (between 20° and 800°C) (m/mK)
A
B
C
1.56 2.68
1.72 2.87
1.72 2.86
4.3 19.8 0.68 77.4 6.22
4.4 19.5 0.27 98.5 6.82
5.9 20.2 0.08 103.5 7.38
316
Tulyaganov, Olhero, Ribeiro, Ferreira, and Labrincha
Fig. 5. SiO2-Al2O3-CaO equilibrium phase diagram indicating the locations of the simplified compositions A, B, and C.
Fig. 6. XRD patterns of powdered samples of compositions A, B, and C after firing at 1500°C for 1.5 h.
Fig. 7. XRD patterns of powdered samples of composition A after firing at different temperatures: 1300°, 1400°, and 1500°C.
Mullite-Alumina Refractory Ceramics Obtained from Mixtures
317
intensity of mullite peaks first tends to increase (1400°C) and then to decrease (1500°C). The following reaction sequences may explain the observed phase evolution: (i) primary mullite (2Al2O3SiO2) is formed from metakaolinite at low temperatures; (ii) this primary mullite then dissociates into the less alumina rich (3:2) final mullite, with needed silica being derived from the continuous decomposition of metakaolinite; (iii) at higher temperatures, mullite dissolution occurs in the increasing amount of liquid phase. A similar reaction sequence was described for metakaolinite-alumina mixtures [23, 24]. SEM observations (Figs. 8 and 9) clearly reveal the presence of an abundant glassy phase involving the crys-
talline grains of samples fired at 1500°C. At this temperature only small microstructural differences could be noticed among samples even between the two extreme compositions A and C. Mullite prismatic grains are visible in highly magnified micrographs, and their abundance is obviously higher in sample A. Closed pores are homogeneously distributed in the ceramic matrix. These pores result from late decomposition reactions of some minor sludge components such as sulphates, and are responsible for the relatively low values of flexural strength. Some of those salts still remain in the sludge even after calcinations at about 1350°C; therefore a careful washing operation of this component is strongly recommended.
Fig. 8. Typical microstructure (SEM) of sample A sintered at 1500°C. Higher-magnified view (b) shows mullite prismatic grains.
Fig. 9. Typical microstructure (SEM) of sample C sintered at 1500°C. Higher-magnified view (b) shows mullite prismatic grains.
318 4. CONCLUSIONS The production of mullite-based refractory ceramics incorporating Al-rich anodizing sludge up to levels of 70 wt.% seems to be functionally viable and interesting from environmental and economical viewpoints. The slip casting technique was optimized and relevant working parameters adjusted to obtain a good control of size and shape of formed bodies. Best results were obtained for a solids batch (wt.%) of 70% sludge 24% kaolin 6% ball clay, and from a suspension containing 55 wt.% solids dispersed with 0.5 wt.% Dolapix. The use of fine-grain diatomite particles reduces the green density of cast bodies, and sintered properties are consequently poorer. In all cases, sludge pretreatment steps such as calcination, washing, and deagglomeration should be performed to eliminate undesirable constituents like soluble salts, to facilitate the mixing process of the components and to improve the sintering process of the ceramic products.
ACKNOWLEDGMENTS Financial support from FCT (Portuguese Foundation for Science and Technology—Project POCTI/CTA/ 42448/01) and from PRODEP 3 is greatly appreciated.
REFERENCES 1. P. Nunes, M. J. Ribeiro, J. M. F. Ferreira, C. S. Bóia and J. A. Labrincha, in Proc. TMS Fall Meet. Recycling Waste Treat. Mineral Metal Process. Tech. Econ. Aspects, B. Bjorkman, C. Samuelsson, J. Wikstrom, eds. (Lulea, Sweden, 2002) vol. 2, pp. 359–368. 2. J. S. Viland, in A Secondary View of Recycling: Recycling of Metals and Engineered Materials (The Minerals, Metals & Materials Society, 1990) p. 427.
Tulyaganov, Olhero, Ribeiro, Ferreira, and Labrincha 3. E. Gonzalez, J. L. Asenjo, A. Baena, J. Dufour, A. La Iglesia, A. Hernadéz, N. Cornejo, E. Ruiz-Ayúcar, E. García, N. Ayala F. García-Carcedo, F. Delmas, and J. R. Mayral, in Proc. Global Symp. Recycl. Waste Treat. Clean Technol. REWAS’99, (TMS and INASMET, 1999) vol. 1, pp. 481– 490. 4. D. A. Pereira, C. M. Bóia, and J. A. Labrincha, Portuguese Patent N°102573, 5th March 2001. 5. M. Pelino, A. Karamanov, C. Cantalini, P. Pisciella, and F. Ullu, in Proc. Global Symp. Recycl. Waste Treat. Clean Technol. REWAS’99, (TMS and INASMET, 1999) vol. 1, pp. 443– 451. 6. M. Dondi, M. Marsigli, and B. Fabbri, Tile Brick Int. 13, 302–308 (1997). 7. D. A. Pereira, J. B. Aguiar, F. P. Castro, M. F. Almeida, and J. A. Labrincha, Cement Concrete Res. 30, 1131–1138 (2000). 8. D. A. Pereira, D. M. Couto, and J. A. Labrincha, CFI Ceram. Forum Int. 77, 21–25 (2000). 9. V. L. Balckevich, Technical Ceramics (Stroiizdat, Moscow, 1984). 10. I. A. Aksay, D. M. Dabbs, and M. Sarikaya, J. Am. Ceram. Soc. 74, 2343–2358 (1991). 11. M. Mizuno, J. Am. Ceram. Soc., 74, 3017–3022 (1991). 12. F. J. Klug, S. Prochazka, and R. Doremus, J. Am. Ceram. Soc. 70, 750–759 (1987). 13. F. Kara, S. Turan, J. A. Little, and K. M. Knowles, J. Am. Ceram. Soc. 83, 369–387 (2000). 14. A. B. Fuentes, J. E. Navarro, and C. A. Jimenez, Bol. Soc. Esp. Ceram. Vidrio 22, 143–148 (1982). 15. R. J. Pugh and L. Bergstrom, in Surface and Colloid Chemistry in Advanced Ceramic Processing: Surfactant Science Series (Marcel Dekker Inc. 1994) Vol. 51. 16. G. A. Parks, Chem. Rev. 65, 177–198 (1965). 17. J. M. F. Ferreira, J. Eur. Ceram. Soc. 18, 1161–1169 (1998). 18. J. M. F. Ferreira and H. M. Diz, J. Am. Ceram. Soc. 82, 1993–2000 (1999). 19. E. F. Osborn and A. Muan, in Phase Equilibrium Diagrams of Oxide Systems (The American Ceramic Society, Ohio, 1960), Plate 3. 20. P. V. Vasconcelos, J. A. Labrincha, and J. M. F. Ferreira, J. Eur. Ceram. Soc. 20, 201–207 (2000). 21. M. D. Sacks, K. Wang, G. W. Sheiffele, and N. Bozkurt, J. Am. Ceram. Soc. 80, 663– 672 (1997). 22. W. Moriano and R. H. Kiminami, in Ceramic Transactions: Ceramic Proceedings Science and Technology, H. Hausner, ed. (The American Ceramic Society, Westerville OH, 1995), vol. 51, pp. 555–559. 23. J. Pascual, J. Zapatero, M. C. Jimenez de Nato, A. J. Ramires del Valle, J. L. Perez-Rodriguez, and P. J. Sanchez-Soto, J. Am. Ceram. Soc. 83, 2677–2680 (2000). 24. K. C. Liu, G. Thomas, A. Caballero, J. S. Moya, and S. de Aza, J. Am. Ceram. Soc. 77, 1545–1552 (1994).
