Evolution of Mullite Texture on Firing Tape-Cast Kaolin Bodies

Chin-Yi Chen and Wei-Hsing Tuan*Institute of Materials Science and Engineering, National Taiwan University, Taipei, Taiwan 106, Republic of China

Kaolin undergoes a series of phase changes on heating toelevated temperature, proceeding through kaolin, metakaolin,�-Al2O3 spinel, and mullite. The morphologic evolution of thekaolin–mullite reaction series was investigated in the presentstudy. A highly textured kaolin green body was prepared by atape-casting technique, and the morphology evolution fromkaolin flakes to mullite aciculars on firing from 450° to 1600°Cwas then monitored using X-ray diffractometry and scanningelectron microscopy. Equiaxed mullite nuclei first appeared at1000°C; the aspect ratio of the mullite grains then increasedwith increased firing temperature. The mullite aciculars re-arranged their orientation in a glassy phase until they im-pinged on each other. A highly textured mullite specimen wasprepared by firing the kaolin tape at a temperature >1500°C.

I. Introduction

THE crystalline product after firing kaolin at a temperature�1100°C is mainly mullite. Although the kaolinite–mullite

reaction series has been extensively investigated for many yearsusing various techniques, such as X-ray diffractometry (XRD),1–15

differential thermal analysis (DTA),8,11,12,14,16–18 and transmis-sion electron microscopy (TEM),3,6,9,10,14,18–23 controversial is-sues continue to exist. These issues include (1) the extent ofcrystallization of metakaolinite, (2) the crystal structure andchemistry of an intermediate, �-Al2O3 spinel phase, which isformed at �1000°C, (3) the composition of mullite at the begin-ning of its formation, and (4) the crystallographic orientationrelationships of the kaolinite–metakaolinite–spinel–mullite series.These issues can be related to the presence of impurities, which areintrinsic in the kaolinite extracted from natural resources. Further-more, the grain size of the crystallized phases involved in thereaction series, such as metakaolinite and �-Al2O3 spinel phase, isat the nanometer scale, making technical analyses challenging.However, because of the development of high-resolution transmis-sion electron microscopy (HR-TEM), considerable progress hasbeen made recently on these issues.14,21

Although several groups have investigated the evolution of thecrystallographic orientation from kaolinite, metakaolinite, andspinel to mullite,5,14,15,19–21 uncertainties on the orientation rela-tionships between kaolin and mullite remain. Brindley and Naka-hira5 and Comer19 proposed that a preferred orientation existsbetween either kaolin and mullite or spinel and mullite. Onike etal.20 observed a relationship between kaolin and spinel. Lee etal.14 confirmed the existence of the preferred orientation betweenspinel and metakaolinite, although their study also suggested thatno clear orientation relationship exists between kaolinite and

mullite. Because the kaolin particles are flaky in nature, McCon-ville et al.21 suggested that the texture development dependsstrongly on the shaping technique used. Textured mullite speci-mens were obtained by preparing kaolin powder compacts by anextrusion21 or a die-pressing technique,15 then fired at elevatedtemperature. However, no mullite texture was observed in a firedloose-packing kaolin powder compact.21

A textured microstructure can offer many possibilities to im-prove the performance of ceramics. For example, mechanical,24

electrical,25 and piezoelectric26 properties can be improved bytransforming an equiaxed microstructure to a textured microstruc-ture. A recent study has demonstrated that a highly textured mullitecan be prepared by templated grain growth,27 specifically, bydoping oriented acicular mullite seeds into a mullite precursor todevelop a textured microstructure after firing at elevated temper-ature. In the present study, a textured mullite has been preparedusing kaolin powder as starting material. The evolution of phasesand their morphologies in the kaolin–mullite reaction series hasbeen carefully monitored.

II. Experimental Procedure

A kaolin powder (AKIMA 35, Akima Co., Malaysia) was usedas the starting material in the present study. Chemical compositionof the kaolin powder was determined using induced coupledplasma emission spectroscopy (ICP; Spectro Analytical Instru-ments, Fitchburg, MA). Kaolin powder and 2 wt% dispersant(sodium hexametaphosphate, Showa Chemical Co., Tokyo, Japan)were mixed together in deionized water in an agitator for 4 h. Thesolids loading was 20 wt%. After the slurry was left unstirred for24 h, the sediment from the slurry was discarded to remove theagglomerates. Approximately 60 wt% of the starting powder wasremoved at this stage. The suspension was then stirred again andheated to 90°C, after which 2.5 wt% binder (poly(vinyl alcohol),PVA; Nacalai Tesque, Inc., Kyoto, Japan) and plasticizer(poly(propylene glycol), PPG; Nacalai Tesque) relative to theslurry were added slowly into the slurry. The ratio of the binder toplasticizer was one to one. The viscosity of the slurry was thenadjusted to 400 cP (0.4 N�s/m2) by removing part of the water byheating at 90°C. Tape casting was conducted after the slurry hadcooled to room temperature and had been sieved.

The slurry was cast onto a moving mylar carrier (24.5 cm wide)through a 15 cm wide doctor blade at a rate of 20 cm/min. Theblade height was set at 400 �m, and the thickness of the green taperanged from 100 to 150 �m after drying. Figure 1 gives thedefinition of the parallel and perpendicular planes of the tape

W. E. Lee—contributing editor

Manuscript No. 188153. Received November 13, 2000; approved February 4,2002.

Supported by the National Science Council, Republic of China, through ContractNo. NSC88-2216-E002-027.

*Member, American Ceramic Society.Fig. 1. Schematic of the casting tape, illustrating the notation used in thetext.

J. Am. Ceram. Soc., 85 [5] 1121–26 (2002)

1121

journal

relative to the casting direction. The cast tape was then stampedinto 1 cm � 1 cm squares and fired at temperatures varying from450° to 1600°C for 1 h after removing the organics at 450°C. Theheating and cooling rates were 5°C/min.

The viscosity of the slurry was measured using a shear-controlled cylinder viscometer (Model LVDT-II�, BrookfieldEngineering Laboratories, Stoughton, MA). The phase identifica-tion was performed using XRD with CuK� radiation. The micro-structure was observed using either SEM or field-emission scan-ning electron microscopy (FE-SEM). Concentrated hydrofluoricacid (48%) was used as etching solution to reveal the morphologyof mullite grains. Many mullite grains were found on the parallelplane, as demonstrated later. Therefore, the size and aspect ratio ofmullite grains could be determined by measuring such grainsdirectly from SEM micrographs. At least six micrographs weretaken from each firing condition, and more than 100 mullite grainswere measured for each firing condition.

III. Results and Discussion

The constituents of the as-received kaolin powder, as deter-mined by ICP, are SiO2 (48.6 wt%), Al2O3 (35.7 wt%), K2O (1.2wt%), Fe2O3 (0.9 wt%), TiO2 (0.4 wt%), MgO (0.2 wt%), CaO(0.1 wt%), BaO (0.1 wt%), and an ignition loss of 12.6 wt%.Comparing these constituents with the chemical formula for

kaolinite, Al2O3�2SiO2�2H2O, there is an excess of �6% SiO2 inthe raw material. The excess SiO2 exists mainly in the form ofquartz in the raw material, as shown in Fig. 2(a). There is also asmall amount of mica impurity detected in the as-received kaolin.Figure 3 shows the morphology of as-received kaolin particles,demonstrating the flaky nature of the particles. Figure 4 shows athree-dimensional view of the green tape, indicating that mostkaolin flakes are aligned parallel. The texture of the green tape canbe further verified using XRD analysis. Figure 2 also shows theXRD patterns of the as-received kaolin powder and of the greentape. The XRD pattern of the as-received kaolin powder isobtained by tapping the powder lightly into the XRD powderholder. Some of the flakes lie flat because of the tapping. Figure

Fig. 2. XRD patterns of (a) as-received powder and (b) green tape.

Fig. 3. Morphology of the kaolin particles.

Fig. 4. Three-dimensional view of a green kaolin tape.

Fig. 5. XRD patterns of the tapes after firing at the indicated tempera-tures for 1 h.

1122 Journal of the American Ceramic Society—Chen and Tuan Vol. 85, No. 5

2(b) shows the XRD pattern of the green tape, demonstrating thepresence of strong peaks of (00l) planes. Kaolinite is a layeredstructure composed of sequential SiO4 tetrahedral sheets andAl(O,OH)6 octahedral sheets.28 The morphology of the kaolinparticles thus tends to be flaky in shape. The c-axis of the kaolinitecrystal is perpendicular to the flake plane. The strong (00l) peaksin Fig. 2(b) indicate that most kaolin flakes are aligned parallel.Thus, the tape-casting technique used in the present study preparesa strongly textured kaolin powder compact. The tape-castingtechnique is superior to the die-pressing technique15 in terms oftexture control of the green compact. Therefore, the morphologicevolution from kaolin flakes to mullite aciculars can be monitored.

The phase transformation series from kaolin to mullite can beseen in the XRD patterns shown in Fig. 5. The kaolinite transformsto metakaolinite by removing the hydroxyl groups at a temperature�600°C. The crystalline structure of metakaolinite is in a form ofshort-range order;4,29 therefore, the diffraction peaks of meta-kaolinite are buried in the XRD background. The size of �-Al2O3

spinel crystals formed after heating at 1000°C is also small, and,therefore, the height of the peaks for the spinel is too low to beidentified. Cristobalite and mullite are found after firing at 1000°Cfor 1 h, suggesting that mullite is formed simultaneously with theformation of spinel. Other groups also have reported this find-ing.14,21 Because the spinel transforms to mullite above 1100°C,

the intensity of the mullite peaks increases significantly afterheating at 1100°C. The intensity of mullite peaks also increaseswith increasing firing temperature. The splitting of (120) and (210)peaks after heating to 1600°C can be attributed to the change of thecomposition from aluminum-rich mullite, 2Al2O3�SiO2, to3Al2O3�2SiO2 mullite.14

The microstructure of the kaolin tape after firing at 1100°C for1 h is shown in Fig. 6. The micrograph for the green tape is alsoshown for comparison purposes. The flaky nature of the startingparticles remains after firing up to 1100°C. From the perpendicularplane, small particles in the range of 0.5–1 �m are found on theedge of the flakes above 1100°C (see Fig. 7(a)). Even smallernuclei, �50 nm, embedded in the particles are revealed afteretching with hydrofluoric acid for a few seconds (see Fig. 7(b)).From XRD analysis (see Fig. 5), these nanosized nuclei aredetermined to be mullite. The mullite nuclei are roughly equiaxed.The morphologies of the mullite grains in the specimens fired at1200° and 1400°C for 1 h are shown in Fig. 8, indicating that themullite grains are no longer equiaxed after firing above 1200°C.Figure 9 shows the morphology of mullite aciculars after firing at1500° and 1600°C for 1 h. The mullite aciculars undergo consid-erable grain growth after heating at 1600°C, and there are manyaciculars aligned parallel, as shown in Figs. 8 and 9. Therefore, thelength, width, and aspect ratio of the mullite aciculars can be

Fig. 6. Morphology of the particles on the parallel plane in (a) green kaolin tape and in (b) the specimens fired at 1100°C for 1 h.

Fig. 7. Morphology of the particles on the perpendicular plane of the specimen fired at 1100°C for 1 h. Microstructure in the square in (a) is shown in (b).Specimen was treated with hydrofluoric acid for a few seconds.

May 2002 Evolution of Mullite Texture on Firing Tape-Cast Kaolin Bodies 1123

determined by measuring such grains in Fig. 10. The points showthe average values, and standard deviation is also shown. Theaspect ratio of mullite aciculars is 10 after firing at 1600°C for 1 h.The standard deviation of the aspect ratio increases with increasingfiring temperature, indicating that the distribution of the aspectratio of aciculars is broadening with increasing temperature. Figure11 shows the three-dimensional view of the tape after firing at1600°C. The three-dimensional micrograph demonstrates astrongly textured microstructure.

The XRD patterns of specimens fired at 1600°C are shown inFig. 12. The XRD pattern of a loose-packing specimen prepared bysimply placing the kaolin powder (without packing or tapping) intoan alumina crucible and then firing at 1600°C is shown in Fig.12(a). Here, the intensity of the mullite (hkl) l 0 peaks in thekaolin tape after firing at 1600°C is significantly lower than that ofthe peaks in the loose-packing specimen. The c-axis of the mulliteis parallel to the long axis of the aciculars.30 The XRD analysis inFig. 12(b) confirms the microstructural observation that manymullite aciculars are aligned in parallel. Although the intensity ofmullite (hkl) l 0 peaks does not decrease to zero, Fig. 11 showsthat those aciculars that are not lying on the parallel plane deviateonly a few degrees from the parallel plane.

The intensity ratio of (100) to (220) planes of mullite can betaken as an index for preferred orientation. The mullite specimenused to determine the intensity ratio shown in the PowderDiffraction File31 is presumably randomly oriented. The index for

Fig. 8. Microstructures of the specimens fired at (a) 1200° and (b) 1400°C for 1 h. Specimens were treated with hydrofluoric acid for a few seconds.

Fig. 9. Microstructures of the specimens fired at (a) 1500° and (b) 1600°C for 1 h. Specimens were treated with hydrofluoric acid for a few seconds.

Fig. 10. Width, length, and aspect ratio of mullite grains as a function offiring temperature. Dwell time was 1 h.

1124 Journal of the American Ceramic Society—Chen and Tuan Vol. 85, No. 5

the preferred orientation for the ratio from the Powder DiffractionFile is thus taken as zero. The index is shown as a function of firingtemperature in Fig. 13. The intensity ratio for the loose-packingspecimen is similar to that taken from the Powder Diffraction File,indicating that the mullite aciculars in the loose-packing specimenare randomly oriented. Figure 7 indicates that the nanosizedmullite nuclei formed at 1100°C are equiaxed. The mullite nucleithen transform to aciculars with an aspect ratio of 2 after firing at1300°C for 1 h, as shown in Fig. 10. Figure 13 suggests that themullite aciculars are randomly oriented at 1300°C. The intensityratio then decreases with increasing firing temperature, indicatingthat the extent of the preferred orientation increases.

A small amount of K2O is present in the starting material. Theeutectic temperature of the K2O–Al2O3–SiO2 system is as low as985°C.32 The mullite nuclei formed at 1100°C are random oriented(see Fig. 7(b)). The presence of the liquid phase facilitates theformation of mullite aciculars.30,33 These mullite grains grow insize within the liquid phase through a dissolution and reprecipita-tion process.30,34 After the formation of mullite aciculars due tothe presence of liquid phase, the aciculars can then accommodatetheir orientation with each other to grow in size. Because kaolinflakes are packed flake by flake, these flakes tend to stick to eachother after being fired at elevated temperature.21 The perpendicular

growth of aciculars in the kaolin tape is constrained because of thelimited space in that direction. The growth of aciculars is,nevertheless, relatively easy parallel to the tape; therefore, atextured mullite is formed after firing the kaolin tape at elevatedtemperature. Furthermore, these mullite aciculars, which are not inparallel and difficult to grow, can dissolve into the liquid and thenprecipitate onto the large aciculars. The extent of preferredorientation, therefore, increases with increasing sintering temper-ature, as shown in Fig. 13. The texture is not noted in theloose-packing specimen, which indicates that the rearrangementprocess is not significantly affected by gravity, but dependsstrongly on the packing of kaolin flakes.

Figure 14 shows that the intensity ratio as a function of firingtime at 1600°C tends to remain constant at that temperature. Thesize and aspect ratio of mullite aciculars are relatively large afterfiring to 1600°C (see Figs. 9(b) and 10). Because the mulliteaciculars can easily impinge on each other, the reorientationprocess is restricted at 1600°C.

IV. Conclusions

The morphologic evolution of the kaolinite–mullite reactionseries is monitored in the present study. A highly textured kaolinpowder compact is first prepared using a tape-casting technique.The kaolin particles are flaky in shape, tending to align parallel tothe tape plane during casting. The flaky nature of the starting

Fig. 11. Three-dimensional view of the specimen fired at 1600°C for 1 h.Specimen was treated with hydrofluoric acid for a few seconds.

Fig. 12. XRD patterns of (a) a loose-packing specimen and (b) kaolintape after firing at 1600°C for 1 h. Peaks with (hkl) l/ 0 in (b) areindicated with arrows.

Fig. 13. Intensity ratio of (001) to (220) planes of mullite as a function offiring temperature. Firing time was 1 h.

Fig. 14. Intensity ratio of (001) to (220) planes of mullite as a function offiring time at 1600°C.

May 2002 Evolution of Mullite Texture on Firing Tape-Cast Kaolin Bodies 1125

particles remains approximately the same up to 1100°C, and themullite nuclei formed at this temperature are equiaxed. Mulliteaciculars formed above 1200°C are caused by the presence of aglassy phase. The mullite aciculars tend to develop parallel to thetape plane because of the space constraint in the perpendiculardirection. The aspect ratio of the mullite aciculars increases withincreasing firing temperature. The reorientation process is prohib-ited, because the mullite aciculars are too large and impinge oneach other. Most mullite aciculars not on the parallel plane deviateonly a few degrees from the parallel plane; thus, the mullitespecimen prepared by firing kaolin tape is textured. The presentstudy demonstrates that there is a preferred orientation betweenkaolin flakes and mullite aciculars, although the extent of preferredorientation depends strongly on the processing conditions.

Acknowledgments

The help from Dr. S. J. Chen, Hocheng Pottery Co., Taiwan, on ICP analysis ishighly appreciated.

References

1F. Rinne, “Rontgenographische Diagnostik beim Brennen von Kalkstein, Dolomit,Kaolin, und Glimmer,” (X-ray Study of Calcined Calcite, Dolomite, Kaolinite, andMica), Z. Kristallogr., 61 [1/2] 113–24 (1924).

2J. F. Hyslop and H. P. Rooksby, “Further Note on Crystalline Break-Up ofKaolin,” Trans. Ceram. Soc. (Engl. Transl.), 27 [4] 299–302 (1928).

3R. Roy, D. M. Roy, and E. E. Francis, “New Data on Thermal Decomposition ofKaolinite and Halloysite,” J. Am. Ceram. Soc., 38 [6] 198–205 (1955).

4G. W. Brindley and M. Nakahira, “The Kaolinite–Mullite Reaction Series: II,Metakaolin,” J. Am. Ceram. Soc., 42 [7] 314–18 (1959).

5G. W. Brindley and M. Nakahira, “The Kaolinite–Mullite Reaction Series: III, TheHigh-Temperature Phases,” J. Am. Ceram. Soc., 42 [7] 319–24 (1959).

6J. J. Comer, “Electron Microscope Studies of Mullite Development in FiredKaolinites,” J. Am. Ceram. Soc., 43 [7] 378–84 (1960).

7A. J. Leonard, “Structural Analysis of the Transition Phases in the Kaolinite–Mullite Thermal Sequence,” J. Am. Ceram. Soc., 60 [1–2] 37–43 (1977).

8A. K. Chakraborty and D. K. Ghosh, “Reexamination of the Kaolinite-to-MulliteReaction Series,” J. Am. Ceram. Soc., 61 [3–4] 170–73 (1978).

9K. Okada, N. Otsuka, and J. Ossaka, “Characterization of Spinel Phase Formed inthe Kaolin–Mullite Thermal Sequence,” J. Am. Ceram. Soc., 69 [10] C-251–C-253(1986).

10B. Sonuparlak, M. Sarikaya, and I. A. Aksay, “Spinel Phase Formation Duringthe 980°C Exothermic Reaction in the Kaolinite-to-Mullite Reaction Series,” J. Am.Ceram. Soc., 70 [11] 837–42 (1987).

11J. A. Pask and A. P. Tomsia, “Formation of Mullite from Sol–Gel Mixtures andKaolinite,” J. Am. Ceram. Soc., 74 [10] 2367–73 (1991).

12A. D. Papargyris and R. D. Cooke, “Structure and Mechanical Properties ofKaolin Based Ceramics,” Br. Ceram. Trans., 95 [3] 107–20 (1996).

13K. J. D. MacKenzie, J. S. Hartman, and K. Okada, “MAS NMR Evidence for thePresence of Silicon in the Alumina Spinel from Thermally Transformed Kaolinite,”J. Am. Ceram. Soc., 79 [11] 2980–82 (1996).

14S. Lee, Y. J. Kim, and H-S. Moon, “Phase Transformation Sequence fromKaolinite to Mullite Investigated by an Energy-Filtering Transmission ElectronMicroscope,” J. Am. Ceram. Soc., 82 [10] 2841–48 (1999).

15C. Y. Chen, G. S. Lan, and W. H. Tuan, “Microstructural Evolution of Mulliteduring the Sintering of Kaolin Powder Compacts,” Ceram. Int., 26 [7] 715–20 (2000).

16P. S. Nicholson and R. M. Fulrath, “Differential Thermal Calorimetric Determi-nation of the Thermodynamic Properties of Kaolinite,” J. Am. Ceram. Soc., 53 [5]237–40 (1970).

17K. J. D. MacKenzie, “Comment on Differential Thermal Calorimetric Determi-nation of the Thermodynamic Properties of Kaolinites,” J. Am. Ceram. Soc., 54 [3]174 (1971).

18K. Srikrishna, G. Thomas, R. Martines, M. P. Corral, S. De Aza, and J. S. Moya,“Kaolinite–Mullite Reaction Series: A TEM Study,” J. Mater. Sci., 25, 607–12(1990).

19J. J. Comer, “New Electron-Optical Data on the Kaolinite–Mullite Transforma-tion,” J. Am. Ceram. Soc., 44 [11] 561–63 (1961).

20F. Onike, G. D. Martin, and A. C. Dunham, “Time–Temperature–TransformationCurves for Kaolinite,” Mater. Sci. Forum, 7 [1] 73–82 (1986).

21C. McConville, W. E. Lee, and J. H. Sharp, “Comparison of MicrostructuralEvolution between Kaolin Powders and Dense Clay Bodies,” Br. Ceram. Proc., 58,75–92 (1998).

22Y. Iqbal and W. E. Lee, “Fired Porcelain Microstructures Revisited,” J. Am.Ceram. Soc., 82 [12] 3584–90 (1999).

23Y. Iqbal and W. E. Lee, “Microstructural Evolution in Triaxial Porcelain,” J. Am.Ceram. Soc., 83 [12] 3121–27 (2000).

24K. Hirao, M. Ohashi, M. E. Brito, and S. Kanzaki, “Processing Strategy forProducing Highly Anisotropic Silicon Nitride,” J. Am. Ceram. Soc., 78 [6] 1687–90(1995).

25T. Takenaka and K. Sakata, “Grain Orientation and Electrical Properties ofHot-Forged Bi4Ti3O12 Ceramics,” Jpn. J. Appl. Phys., 19 [1] 31–39 (1980).

26H. Igarashi, K. Matsunaga, T. Taniai, and K. Okazaki, “Dielectric and Piezo-electric Properties of Grain-Oriented PbBi2Nb2O9 Ceramics,” Am. Ceram. Soc. Bull.,57 [9] 815–17 (1978).

27S-H. Hong and G. L. Messing, “Development of Textured Mullite by TemplatedGrain Growth,” J. Am. Ceram. Soc., 82 [4] 867–72 (1999).

28J. W. Gruner, “Crystal Structure of Kaolinite,” Z. Kristallogr., 83 [1–2] 75–88(1932).

29G. W. Brindley and M. Nakahira, “The Kaolinite–Mullite Reaction Series: I, ASurvey of Outstanding Problems,” J. Am. Ceram. Soc., 42 [7] 311–14 (1959).

30S. M. Johnson and J. A. Pask, “Role of Impurities on Formation of Mullite fromKaolinite and Al2O3–SiO2 Mixtures,” Am. Ceram. Soc. Bull., 61 [8] 838–42 (1982).

31Powder Diffraction File No. 15–776, International Centre for Diffraction Data,Newtown Square, PA, 1995.

32E. M. Levin, C. R. Robbins, and H. F. McMurdie, Fig. No. 407 in PhaseDiagrams for Ceramists, Vol. 1. Edited by M. K. Reser. American Ceramic Society,Columbus, OH, 1964.

33H. Song and R. L. Coble, “Origin and Growth of Platelike Abnormal Grains inLiquid-Phase-Sintered Alumina,” J. Am. Ceram. Soc., 73 [7] 2077–85 (1990).

34W. M. Carty and U. Senapati, “Porcelain—Raw Materials, Processing, PhaseEvolution, and Mechanical Behavior,” J. Am. Ceram. Soc., 81 [1] 3–20 (1998). �

1126 Journal of the American Ceramic Society—Chen and Tuan Vol. 85, No. 5