Effect of Fluoride and Oxide Additives on the Phase Transformations in Alumina/Clay Ceramics

N. Tezuka 1*, I. M. Low 1, I. J. Davies 2, I. D. Alecu 3, R. J. Stead 3, E. G. Mehrtens 4 and B. A. Latella 4

Department of Applied Physics, Curtin University of Technology, GPO Box U1987, Perth, WA 6845, Australia Department of Mechanical Engineering, Curtin University of Technology, GPO Box U1987, Perth, WA 6845, Australia 3 Rojan Advanced Ceramics Pty Ltd., PO Box 7126, Spearwood, WA 6163, Australia 4 Institute of Materials and Engineering Science, Australian Nuclear Science and Technology Organisation, Private Mail Bag 1, Menai, NSW 2234, Australia
2 1

*Email: nobuo.tezuka@postgrad.curtin.edu.au Available Online at: www.austceram.com/ACS-Journal-2009vol1.asp

Abstract The effects of various types of fluorides (calcium fluoride (CaF2), sodium fluoride (NaF) and potassium fluoride (KF)) on physical and mechanical properties, and oxides (titanium dioxide (TiO2) and ferric oxide (Fe2O3)) on mullitisation, in alumina/clay ceramics were investigated. Two different types of kaolin clays (Al2Si2O5(OH)4), i.e., kaolinite and halloysite, were utilised, although kaolinite mixed with the fluoride additives was dehydroxylated in advance. The fluorides generally improved sinterability of the alumina/kaolinite ceramics at temperatures approaching 1300C due to liquid phase sintering, but higher sintering temperatures and excess amounts of fluorides reduced the properties due to the presence of sealed porosity and glassy phases. The oxide-containing sintered samples were quantitatively analysed by X-ray diffraction (XRD) with an internal standard using the Rietveld method. The major phase was found to be mullite with the Fe3+ ions tending to incorporate into the mullite structure more than that of the Ti4+. Keywords: Mullite, Additives, Fluoride, Clay, Strength, Transformation.

INTRODUCTION Monolithic mullite (3Al2O32SiO2) has been studied and used for various applications due to its superior mechanical, thermal, chemical and electrical properties [1]. Mullite has been mainly prepared using the following methods [2]: (1) mixtures of sols and/or oxides, (2) coprecipitated gels, (3) clay minerals and (4) glassy states. Out of these, the sol-gel method has been most commonly used for the production of chemically pure mullite. However, clay has also been widely used, not only for reasons of economy, but also due to its potential for the formation of needle-like mullite grains [3] which is preferable from the point of view of high strength and toughness. Kaolinite, which belongs to the kaolin group (Al2Si2O5(OH)42H2O), has perhaps been the most popular clay in this research field, with other clays of interest including halloysite (kaolin group), kyanite, etc. Fine mullite grains may be obtained from halloysite following sintering [4], however the authors have observed problems such as low densification and significant cracking due to the coexisting cristobalite which might have limited research into this clay. Despite its excellent properties, the widespread usage of mullite has been restricted due to the difficulty of sintering mullite in its pure form

from powder [5]. In response to this, mullite composites containing alumina (Al2O3) and zirconia (ZrO2) have also been investigated [6-9] as these materials may be easily fabricated by sol-gel or reaction sintering which have the advantage of requiring a lower sintering temperature. However, the presence of dissociation at higher temperatures (e.g., 1400 oC) during the reaction sintering of mullite from alumina/clay is known to result in the formation of significant porosity within the composite matrix [10]. Hence, additives such as titanium dioxide (TiO2) [11], ferrous oxide (Fe2O3) [12] and magnesium oxide (MgO) [13] have been utilised in an attempt to reduce the sintering temperature. Whilst oxide-based compounds have been the most widely utilized sintering aids, attempts have also been made to incorporate fluoride-based sintering aids. Compounds have also been utilised with topaz (Al2SiO4(F,OH)2) being considered as an intermediate compound for the formation of mullite [14-16]. The aim of this work is to investigate the effect of fluoride and oxide additives on the formation and resulting properties of sintered alumina/clay ceramics.

19

N. Tezuka , I. M. Low, I. J. Davies , I. D. Alecu , R. J. Stead , E. G. Mehrtens and B. A. Latella

Physical and mechanical properties were measured for alumina/kaolinite composites containing calcium fluoride (CaF2), lithium fluoride (LiF), sodium fluoride (NaF) or potassium fluoride (KF) as a function of sintering temperature. In addition, powder X-ray diffraction (XRD) analysis was conducted for alumina/kaolinite or alumina/halloysite composites containing oxide additives (TiO2 and Fe2O3). The difference in sintering and densification behaviour due to the presence of the various additives was also investigated.

Table 1: Chemical composition of the clay minerals (wt%).

Kaolinite 46.8 37.0 1.01 0.29 0.24 0.8 0.12 0.21 13.45 Halloysite 49.1 35.9 0.29 trace trace 0.07 trace trace 14.08

METHODS AND PROCEDURES Sample preparation Kaolinite (Unimin Australia Ltd.) and halloysite (Imerys Tableware Asia Ltd., New Zealand) were utilized as the starting clay materials for this investigation with their respective chemical compositions being shown in Table 1. The major differences between these clays were: (i) kaolinite contained a higher proportion of impurities such as TiO2 and Fe2O3 and (ii) according to the manufacturers data sheets and qualitative powder XRD (unpublished), halloysite contained a higher amount of cristobalite (SiO2). As the high water content of clay (typically 10-20 wt%) often results in the formation of cracks during sintering, the samples containing fluoride additives were prepared using meta-kaolinite which was dehydrated and dehydroxylated at 750 C prior to mixing the raw materials. Four types of fluoride sintering aids (i.e., CaF2, LiF, NaF and KF) were individually added (1.5 or 3 wt%) to a mixture of 30 wt% meta-kaolinite and 70 wt% alumina in order to examine the effect of fluoride additives on the sinterability of alumina/clay composites. In contrast to this, the oxide additives (1.5 or 3 wt% TiO2; 2.5 or 5 wt% Fe2O3) were added to a mixture of 50 wt% kaolinite (or halloysite) and 50 wt% alumina in order to examine the formation of mullite. In both the fluoride and oxide cases, reference specimens were also produced using no additives. All of the powder mixtures were dry mixed using a tungsten ring mill until the particle size below 1m. Pure alumina specimens were also prepared as the controls for the fluoride series. The fluoride-containing powders were shaped into rectangular bars of approximate size 12 x 60 x 3 mm and then uniaxially pressed at 110 MPa and then cold isostatically pressed (CIP) at 200 MPa prior to sintering at either 1300, 1450 or 1600 C for 10 hrs. with the heating rate of 1.5 Cmin-1 in air. In contrast to this, the oxide-containing powders were CIPed at 120 MPa after uniaxial pressing (113 MPa) into cylindrical rods, then sintered at 1550C for 2 hrs. with the heating rate of 3 Cmin-1 in air.

SiO2 Al2O3 Fe2O3 CaO MgO TiO2 Na2O K2O LOI

Characterization Bulk density (Db) and apparent porosity (Pa) were measured using the Archimedes method for each composition. The fluoride-containing sintered bar samples were initially machined and flattened as necessary. Following this, the tensile surface (during mechanical testing) of each bar specimen was polished down to a 10 micron finish prior to mechanical testing. The dynamic Youngs modulus (Ef) was determined using impulse excitation and the vibration method. Unless the shear modulus was measurable, the Poissons ratio was assumed to be 0.22 (= typical value for alumina) in order to calculate the Youngs modulus. Four-point bend tests were mainly performed using outer and inner spans of 40 and 20 mm, respectively, although for some cases, values of 20 and 10 mm were used when the bar was not sufficiently long. The fourpoint bend tests were conducted using a crosshead speed of 0.3 mmmin-1 on a digital-controlled testing machine (Model 8562, Instron, Canton, MA). Phase evolution within the oxide-containing sintered composites was investigated using powder X-ray diffraction analysis (Model D500, Siemens) using Cu K radiation and a step size of 0.04, a scanning speed of 1.2min-1 and a scanning range of 2 from 10 to 120. The resulting peaks were indexed against Powder Diffraction File (PDF) cards whilst each powder specimen contained 20 wt% of zinc oxide (ZnO) as an internal standard. The Rietveld method [17] was used, in conjunction with the RIETICA software package [18], to quantify the phase abundance within sintered specimens. The mullite model reported by Angel and Prewitt [19] was applied to the refinement.

20

Journal of the Australian Ceramic Society Volume 45[1], 2009, 19-27

4.0 Bulk Density (gcm )

RESULTS AND DISCUSSION Sinterability of alumina/kaolinite containing fluoride additives The influence of fluorides on: (i) the bulk density and (ii) apparent porosity of alumina/kaolinite composites as a function of sintering temperature have been presented in Fig.1 and 2, respectively. The bulk density tended to increase with fluoride content and temperature, whilst the apparent porosity showed the opposite behaviour. Note that, as the phase compositions of the sintered specimens have not been decided at this stage, it is difficult to estimate the true solid density. However, when compared with the calculated theoretical density of the derived alumina/mullite mixture (3.48 gcm-3) assuming densities of 3.98 gcm-3 and 3.12 gcm-3 for alumina and mullite, respectively, then it is clear that the density values shown in Fig.1 are still low even when the apparent porosity is close to zero. A possible explanation for this limitation of densification through the addition of fluoride sintering aids is an increase in sealed porosity and/or significant amount of glassy phase formation. The correlation between bulk density and apparent porosity for these specimens has been shown in Fig.3. The slope was essentially similar for both alumina/clay and pure alumina, but the variation in the slope became larger for the alumina/clay specimens at lower apparent porosity and, in particular, at higher temperatures. In this case, the presence of sealed porosity and/or glassy phases was considered to be the main cause of this variation. Physical properties of the fluorides utilized in this work have been shown in Table 2 [20-22]. Two major aspects may be noted in Table 2 as follows: (i) melting point, which may affect the eutectic temperature of the matrix, and (ii) single bond strength, which determines whether the fluoride acts as a glass former or modifier. Comparing the results in Figures 1 and 2, CaF2 was noted to show a very mild influence on the composite sinterability, most likely due to its relatively high melting point; it possessed only a minor effect in lowering the sintering temperature within the range of this study as also shown in the mechanical properties later. On the other hand, LiF has the lowest melting point in Table 2 and, therefore, a liquid phase is expected to form at a much lower temperature. In fact, the authors observed a glazing-like phenomenon on the LiF samples at 1450C which was not found in other fluoride specimens. However, the bulk density of the LiF specimens was not necessarily higher when compared to those of NaF or KF. This result was attributed to the formation of -spodumene (true density 2.4 gcm-3) as shown by XRD (Fig.4) and the increase in sealed porosity.

(a) Alumina/kaolinite +CaF2

-3

3.5 3.0 2.5 2.0 4.0 0 1 2 (b) Alumina/kaolinite +NaF CaF2 (wt%)

1300C 1450C 1600C

Bulk Density (gcm )

3.5 3.0 2.5 2.0 4.0 0 1 2 (c) Alumina/kaolinite +KF NaF (wt%)

3 1300C 1450C 1600C

-3

Bulk Density (gcm )

-3

3.5 3.0 2.5 2.0 4.0 0 1 2 (d) Alumina/kaolinite +LiF KF (wt%) 3 1300C 1450C 1300C 1450C 1600C

Bulk Density (gcm )

-3

3.5 3.0 2.5 2.0 0 1 2 Additives (wt%)

Fig. 1: Effect of additives on bulk density as a function of sintering temperature for the specimens containing fluoride-based sintering aids.

Table 2 : Physical properties of additives

Melting Boiling point Fluoride point (C) or oxide (C) LiF 848 1676 NaF 996 1695 KF 858 1502 1418 2533 CaF2 1800 TiO2 Fe2O3 1565 Ref. [20] [20] Cation ionic Single bond strength radius (nm) (kcalmol-1) 0.090 34 0.116 30 0.152 29 0.114 46 0.075 0.069 [21] [22]

21

N. Tezuka , I. M. Low, I. J. Davies , I. D. Alecu , R. J. Stead , E. G. Mehrtens and B. A. Latella

40 Apparent Porosity (%)

Intensity (a.u.)

(a) Alumina/kaolinite +CaF2 30 20 10 0 40 0 1300C 1450C 1600C

(a)
1600C M 1450C C

Z Z A ZM MM M A A M M AM M

A/MK Z A

1300C

Intensity (a.u.)

Apparent Porosity (%)

1 2 (b) Alumina/kaolinite +NaF CaF2 (wt%)

30 20 10 0 40 0 1 2 (c) Alumina/kaolinite +KF NaF (wt%)

3 1300C 1450C 1600C

(b)
A 3%, 1450C 3%, 1300C 1.5%, 1300C M

Z Z A Z A

A/MK + KF A A Z

Intensity (a.u.)

(c)
3%, 1300C S S 1.5%, 1300C

A/MK + LiF A S Z Z S M ZA A A Z S A

3 1300C 1450C 1600C

Apparent Porosity (%)

30 20 10 0 40 0

10

20

30

40

50

2 theta ()
1 2 (d) Alumina/kaolinite +LiF KF (wt%) 3 1300C 1450C

Apparent Porosity (%)

30 20 10 0 0 1 2 Additives (wt%)

Fig.4: XRD patterns of alumina-metakaolinite specimens sintered at various temperature and with various additives. (a) non-additive, (b) KF, (c) LiF. (A: corundum, M: mullite, C: cristobalite, S: spodumene, Z: ZnO) NaF and KF were found to improve the bulk density at 1300 C, but a sintering condition of 1600C was found to be too high for these specimens (in terms of densification), in particular for NaF, and this was again attributed to the presence of sealed porosity and/or glassy phase. According to the XRD results for the KF specimens in Fig.4, mullite tends to disappear by KF addition whilst alumina remains strong peaks. Unlike some cations such as Fe3+ and Ti4+ in the oxide additives, the ions of Li, Na, K and Ca are significantly larger than those of Al and Si and, therefore, these cations cannot substitute into mullite but instead may remain in the glassy phase as the modifier, or form another aluminosilicate compounds like -spodumene. Also, Na and K are considered to enhance a phase separation in Al2O3-SiO2 system which results in the evolution of alumina and silica-rich glassy phase. The phase abundance shown in Table 3 which was determined by the quantitative XRD would further support the above explanations.

Fig.2: Effect of additives on apparent porosity as a function of sintering temperature for the specimens containing fluoride-based sintering aids.

35 Apparent Porosity (%) 30 25 20 15 10 5 0 0 1 2 3

-3

Alumina/kaolinite

Alumina

1300C 1450C 1600C

Bulk Density (gcm )

Fig.3: Correlation of bulk density and apparent porosity.

22

Journal of the Australian Ceramic Society Volume 45[1], 2009, 19-27

Table 3: Phase abundance of sintered specimen derived from alumina-metakaolinite with fluoride additives
Phase abundance (wt%) Additives None None None KF: 1.5% KF: 3% KF: 3% LiF: 1.5% LiF: 3% Sintering temp. 1300C 1450C 1600C 1300C 1300C 1450C 1300C 1300C Alumina 67.4 1.3 43.4 1.0 36.1 0.9 71.8 1.7 74.1 1.7 72.0 1.8 72.8 1.8 73.0 1.6 Mullite 13.5 0.4 50.3 1.2 57.4 1.3 10.5 0.8 3.9 0.4 5.4 0.4 6.1 0.5 Quartz 0.4 0.1 0.2 0.1 0.3 0.1 0.1 0.1 0.3 0.1 0.2 0.1 Cristobalite Amorphous Spodumene 4.7 0.2 14.1 1.9 12.9 0.4 20.5 0.6 6.3 2.1 6.4 2.2 17.6 2.5 21.8 2.0 22.5 2.2 7.9 2.3 6.3 2.2

Figure 5 illustrates the influence of fluorides on Youngs modulus (Ef) at various sintering temperatures. The LiF, NaF and KF sintering aids were found to improved Youngs modulus at 1300C when compared to the case of the specimens containing either CaF2 or else no additive. However, the difference between these two groups was noted to decrease as the sintering temperature increased. A primary factor which is known to influence Youngs modulus is the bulk density or apparent porosity as shown in Figure 6. In general, the relationship between porosity and Youngs modulus may be expressed as follows [23]: Ef = E0exp(-bP) (1)

250 Young's Modulus (GPa) (a) Alumina/kaolinite +CaF2 200 150 100 1300C 50 0 250 0 Young's Modulus (GPa) 200 150 100 50 0 250 0 Young's Modulus (GPa) 200 150 100 50 0 250 0 Young's Modulus (GPa) 200 150 100 50 0 0 1 2 Additives (wt%) 3 1300C 1450C 1300C 1450C 1600C 1 2 (d) Alumina/kaolinite +LiF KF (wt%) 3 1 2 (c) Alumina/kaolinite +KF NaF (wt%) 3 1300C 1450C 1600C 1450C 1600C 1 2 (b) Alumina/kaolinite +NaF CaF2 (wt%) 3

where P is the volume fraction porosity, E0 is Youngs modulus when P=0 and b is constant. As the percentage of sealed porosity was unknown in this study, it was not possible to apply Equation 1 to Figure 6(b), in particular for cases where the apparent porosity is low, i.e., the influence of sealed porosity is even larger. However, assuming that the bulk density most likely correlates with the total porosity (provided that the true solid density is similar for each specimen), then Youngs modulus and the bulk density would be expected to posses a consistent correlation, as was indeed shown to be the case in Figure 6(a). The influence of fluoride additives on flexural strength has been presented in Figure 7 with a similar trend being noted as for the Youngs modulus data (although in this case the degree of uncertainty was much larger). Although not shown here, the correlation between flexural strength and bulk density or apparent porosity also exhibited additional scatter when compared to the Youngs modulus data. The relationship between flexural strength (f) and porosity may be expressed as follows [24]:

f =142,500 exp(-11.83P)d(3.33P-0.60)

(2)

where P is the volume fraction porosity and d is the grain size.

Fig.5: Effect of additives on Youngs Modulus as a function of sintering temperature for the specimens containing fluoride-based sintering aids.

23

N. Tezuka , I. M. Low, I. J. Davies , I. D. Alecu , R. J. Stead , E. G. Mehrtens and B. A. Latella

400 1300C 350 Young's Modulus (GPa) 300 250 200 150 100 50 0 0 1 2 3
-3

(a)

1000 Alumina

(b)

1450C 1600C Young's Modulus (GPa)

Alumina

100

Whilst the refinement model for mullite did not take into account the solid solution of titanium and iron ions within mullite, the cell parameters were found to increase with the amount of additive. The relationship between a-axis and cell volume shown in Fig. 10 indicated similar results for both kaolinite and halloysite as that obtained from bauxite by Schneider and Wohlleben [25]. This result suggested that TiO2 and Fe2O3 additives influenced the structure of mullite similarly despite the different starting minerals.

1300C 1450C 1600C 10 4 5 0 20 Pa (% ) 40 Db (gcm )

300 Flexural Strength (MPa) (a) Alumina/kaolinite +CaF2 1300C 200 1450C 1600C 100

Fig.6: Correlation of Youngs Modulus and (a) bulk density,Db and (b) apparent porosity, Pa When the densification is good (i.e., P0), the grain size becomes more crucial to improving the flexural strength. In order to maintain a fine grain size, pinning of the grain boundary by additives such as MgO used in alumina is widely performed. In addition, reducing the sintering temperature and/or time can also prevent abnormal grain growth. As shown in Fig. 7, the highest flexural strengths were obtained in the alumina/kaolinite specimens containing LiF or KF sintered at 1300 C. Thus, the data obtained in this works suggest the possibility of optimal flexural strength data being obtained at relatively low sintering temperatures. In light of these results, the authors suggest that the influence of shorter sintering times at 1300 oC and/or reduced sintering temperature on flexural strength may be worthy of further investigation. Effect of oxide additives on mullitisation in alumina/kaolinite or halloysite As several oxide additives (e.g., Fe2O3, TiO2, MgO, etc.) have been investigated by other researchers, in this work the influence of TiO2 and Fe2O3 on mullite derived from different clays, i.e., kaolinite and halloysite, was examined. XRD analysis indicated mullite to be the dominant phase in all of the oxide specimens investigated as shown in Fig. 8. Minor phases such as corundum, cristobalite, quartz and rutile were also found in several of the specimens. Quantitative analysis of specimens was conducted using the Rietveld method with the amount of glassy (amorphous) phase being determined according to a ZnO internal standard (Fig. 9). The amount of glassy phase was noted to generally increase with the addition of TiO2 or Fe2O3. When the level of additive was low (<2 wt%), the amount of glassy phase increased slightly. However, further increases in the amount of additive resulted in the formation of additional glassy phase.

0 300 0 Flexural Strength (MPa)

1 2 (b) Alumina/kaolinite +NaF CaF2 (wt%)

3 1300C 1450C 1600C

200

100

0 300 0 Flexural Strength (MPa)

200

1 2 (c) Alumina/kaolinite +KF NaF (wt%) 1300C 1450C 1600C

100

0 300 0 Flexural Strength (MPa) 1 2 (d) Alumina/kaolinite +LiF KF (wt%) 3 1300C 1450C

200

100

0 0 1 2 Additives (wt%) 3

Fig.7: Effect of additives on flexural strength as a function of sintering temperature for the specimens containing fluoride-based sintering aids.

24

Journal of the Australian Ceramic Society Volume 45[1], 2009, 19-27

Furthermore, the SiO2 content in mullite was determined by calculating the site occupancy of Si in mullite after refinement. As shown in Fig. 11, TiO2 did not change the SiO2 content significantly for TiO2 amounts of 1.5 wt% or more. On the other hand, the Fe2O3 additive was noted to significantly influence the amount of SiO2 up to the maximum of 5 wt% utilized in this work; this result implied that Fe3+ substituted with Si4+ in the mullite tetrahedral site directly, or else that Fe3+ promoted Al3+ ions from their octahedral site to tetrahedral sites where Si4+ was substituted by Al3+ as reported for Ti4+ by others [26].

100 (a)
Mullite content (wt%)

Kaolinite + TiO2 Halloysite + TiO2 Halloysite + Fe2O3

98 96 94 92 90 10 8 6 4 2 0 0 1 2 3 4 5 6 Additives, TiO2 or Fe2 O3 (wt% )

Glassy phase content (wt%)

(b)

(a)Alumina/kaolinite Intensity (a.u.)

+ TiO2 3wt%

1.5wt% 0wt% (b)Alumina/halloysite + TiO2 3wt% 1.5wt% 0wt% (c)Alumina/halloysite + Fe2 O3 3wt%

Intensity (a.u.)

Fig.9: Effect of additives on (a) mullite and (b) glassy phase content for specimens sintered with oxide additives at 1550 C for 2 hrs (specimens include alumina).
169.0 168.5

Intensity (a.u.)

0wt% mullite 84-1205

Cell volume ( )

1.5wt%

168.0 167.5 167.0 166.5 7.53

Kaolinite + TiO2 Halloysite + TiO2 Halloysite + Fe2O3 Ref. [25]

corundum 83-2080

quartz low 88-2487 alpha cristobalite 89-3434

7.55

7.57

7.59

a-axis ()
Fig.10: Relationship between cell volume and a-axis for specimens sintered with oxide additives at 1550 C for 2 hrs (specimens include alumina).

ZnO 89-1397

15

20 25

30 35 40 2 Theta ()

45

50

55

Fig.8: XRD patterns of alumina/clay after sintering with oxide additives at 1550 C for 2 hrs.

25

N. Tezuka , I. M. Low, I. J. Davies , I. D. Alecu , R. J. Stead , E. G. Mehrtens and B. A. Latella

40.0

SiO2 content in mullite (mol%)

39.5 39.0 38.5 38.0 37.5 37.0 36.5 36.0 35.5 0 1 2 3 4 5 6 Kaolinite + TiO2 Halloysite + TiO2 Halloysite + Fe2O3

REFERENCES:
1. 2. 3. S. Somiya and Y. Hirata, Mullite Powder Technology and Applications in Japan, Am. Ceram. Soc. Bull., Vol. [70], 10, (1991), 1624-1632. K. Okada, N. Otsuka and S. Somiya, Review of Mullite Synthesis Routes in Japan, Am. Ceram. Soc. Bull., Vol. [70], 10, (1991), 1633-1640. C.Y. Chen, G.S. Lan and W.H. Tuan, Microstructural evolution of mullite during the sintering of kaolin powder compacts, Ceram. Int., Vol. [26], 7, (2000), 715-720. D.S. Perera and G. Allott, Mullite Morphology in Fired Kaolinite/halloysite Clays, J. Mater. Sci. Lett., Vol. [4], (1985), 1270-1272. A.J. Skoog and R.E. Moore, Refractory of the Past for the Future: Mullite and Its Use as a Bonding Phase, Am. Ceram. Soc. Bull., Vol. [67], 7, (1988), 1180-1185. P. Miranzo, et al., TEM study of reaction-sintered zirconia-mullite composites with CaO and MgO additions, J. Mater. Sci., Vol. [22], (1987), 2987-2992. A.C. Mazzei and J.A. Rodrigues, Alumina-mullitezirconia composites obtained by reaction sintering: Part I. Microstructure and mechanical behaviour, J. Mater. Sci., Vol. [35], (2000), 2807-2814. T. Nagano, H. Kato and F. Wakai, Diffusion bonding of ceramics: mullite, ZrO2-toughened mullite, J. Mater. Sci., Vol. [26], (1991), 4985-4990. T. Koyama, et al., "Preparation and characterization of mullite-zirconia composites from various starting materials", J. Eur. Ceram. Soc., Vol. [14], 4, (1994), 295-302. C.Y. Chen, G.S. Lan, and W.H. Tuan, "Preparation of mullite by the reaction sintering of kaolinite and alumina", J. Eur. Ceram. Soc., Vol. [20], 14-15, (2000), 2519-2525. H.S. Tripathi and G. Banerjee, "Synthesis and mechanical properties of mullite from beach sand sillimanite: effect of TiO2", J. Eur. Ceram. Soc., Vol. [18], 14, (1998), 2081-2087. H.S. Tripathi, S. Kr Das, and G. Banerjee, "Thermal shock behaviour of high alumina aggregates derived from sillimanite beach sand with and without Fe2O3 doping", Ceram. Int., Vol. [26], 1, (2000), 1-6. L. Montanaro, et al., "Sintering of industrial mullites", J. Eur. Ceram. Soc., Vol. [17], 14, (1997), 1715-1723. H. Zipkin, et al., "Dielectric properties of sodium fluoride added kaolinite at different firing temperatures", Ceram. Int., Vol. [33], 4, (2007), 663-667. K. Okada and N. Otuska, "Synthesis of Mullite Whiskers and Their Application in Composites", J.Am. Ceram. Soc., Vol. [74], 10, (1991), 2414-2418. T.V. Vakalova, A.V. Ivanchenkov and V.M. Pogrebenkov, Kinetics of the Solid-Phase Synthesis of Mullite in the Presence of Topaz, Refractories and Industrial Ceramics, Vol. [46], 1, (2005), 49-55. H.M. Rietveld, A profile refinement method for nuclear and magnetic structures, J. Appl. Cryst., Vol. [2], (1969), 65-71. B.A. Hunter and C.J. Howard, A Computer Program for Rietveld Analysis of X-ray and Neutron Powder Diffraction Patterns, Lucas Heights Research Laboratories, Australian Nuclear Science and Technology Organization, February (2000). R.J. Angel and C.T. Prewitt, Crystal Structure of Mullite: A Re-examination of the Average Structure, Am. Mineral., Vol. [71], (1986), 1476-1482.

4. 5. 6. 7.

Additives (wt%)

Fig.11: Effect of additives on SiO2 content in mullite for specimens sintered with oxide additives at 1550 C for 2 hrs (specimens include alumina).

CONCLUSIONS In this work the influence of fluoride and oxide additives on the mechanical and physical properties (fluorides) and formation of mullite (oxides) in alumina/kaolinite/halloysite composites was investigated 1. All of the fluorides examined in this study (CaF2, LiF, NaF and KF) were found to have contributed to liquid phase sintering with the extent of the contribution depending on the sintering temperature. When the temperature was too high (i.e., 1600 oC), a significant amount of liquid phase was believed to have been introduced, which resulted in the formation of additional sealed porosity and/or glassy phase in the matrix, and consequently lead to a decrease in the bulk density. 2. Youngs modulus exhibited a good correlation with bulk density, whilst the correlation of flexural strength with bulk density was controlled by combination of other factors. The presence of fluoride additives in alumina/kaolinite composite specimens was found to allow a decrease of 300 o C in sintering temperature for the same level of Youngs modulus and flexural strength. 3. Whilst the mullite refinement model did not contain either Fe or Ti, the change of Si amount within the mullite implied that Fe3+ tended to substitute Si4+ in tetrahedral sites, either directly or indirectly, whilst Ti4+ was considered to have been incorporate into octahedral sites.

8. 9.

10.

11.

12.

13. 14. 15. 16.

17. 18.

ACKNOWLEDGMENTS This work was supported by the Australian Institute of Nuclear Science and Engineering (AINSE Award No: 06112). The authors gratefully acknowledge the help of Dr. L. Morales of the Bragg Institute for her help with sintering some of the specimens.

19.

26

Journal of the Australian Ceramic Society Volume 45[1], 2009, 19-27

20. M. Winter, WebElements Periodic Table, The University of Sheffield and WebElements Ltd, UK, <http://www.webelements.com/> at 25 April 2007. 21. A. Nakahara, et al., Mukikagakujosetsu, Kagakudoujin, Kyoto, (1985) [in Japanese] 22. C.M. Baldwin and J.D. Mackenzie, "Fundamental Condition for Glass Formation in Fluoride Systems", J. Am. Ceram. Soc., Vol. [62], 9-10, (1979), 537-538. 23. R.M. Spriggs, "Expression for Effect of Porosity on Elastic Modulus of Polycrystalline Refractory Materials, Particularly Aluminum Oxide", J. Am. Ceram. Soc., Vol. [44], 12, (1961), 628-629. 24. E.M. Passmore, R.M. Spriggs, and T. Vasilos, "Strength-Grain Size-Porosity Relations in Alumina", J. Am. Ceram. Soc., Vol. [48], 1, (1965), 1-8. 25. H. Schneider and K. Wohlleben, "Microchemical composition and cell dimensions of mullites from refractory-grade South American bauxites", Ceram. Int., Vol. [7], 4, (1981), 130-136. 26. E. Tkalec, D. Navala and M. osi, Distribution of titanium and aluminium in sintered mullite, J. Mater. Sci., Vol. [25], (1990), 1816-1820.

27