journal Am Ceram Soc852002) Nucleation and Growth of Mullite Whiskers from Lanthanum-Doped Aluminosilicate Melts Inacio Regiani, Washington L. E. Magalhaes, Dulcina. Ferreira de souza, Carlos O. Paiva-Santos, and Milton Ferreira de souza of Sao Paulo(USP) ersity of Sao Carlos iraquara Campus NESP.SP,Brazil f REAS er 2. 2000: approved August 14, 2001. ial institution FAPESP
200 Nucleation and Growth of Mullite Whiskers from Lanthanum- Doped Aluminosilicate Melts Before choosing the(La, O,)003(3A1203: 2SiO2)097 composi- The results are presented in three parts: the effects caused by the tion, which represents a very small region of the phase diagram, to glass melt below 1500"C and from 15000-1700C, and the process perform the major part of this study, experiments were also of whisker nucleation and growth conducted with most of the rare earths, from Ce to Yb as well as La and Y. These experiments demonstrated that all rare earths (1) Firing below 1500C induce nucleation and growth of mullite whiskers in a similar The DTA analysis of pure and doped compacted mullite powders(Fig. 1) shows well-defined exothermic peaks at 1346. 1240°,1221°,1213°,and1217C, respectively, for the pure and IL. Experimental Procedure lanthania-, neodymia-, erbia-, and yttria-doped compositions, al attributed to the full transformation of the alumina-silica powder Diphasic gels were processed to produce powders with a to the mullite phase. composition of 3.0 mol% lanthana(RE2O3), 58.2 mol alumina The mullite transformation temperature of the LaAs systen (1240C) is very close to the eutectic temperature reported by other compositions, e.g, with an AL, O,/SiO, composition in the Kolitsch" for this system. However, for the other REAS, the 10-3.0 interval and with a concentration of 5.0 mol% lanthana. temperature for mullite transformation, as defined through the he detailed procedure for powder preparation and pellet confor- DTA analysis and confirmed by the XRD pat atterns, is well below mation has been published elsewhere. Pellets were fired for the respective eutectic temperatures reported in the literature. It is well known that a small fraction of liquid phase decreases the the cooling rate used was 1000 C/h. A differential thermal analysis mullitization temperature. The volume of Al O, and Sio, involved (DTA)(Model STA 409C, Netzsch, Selb, Germany)was made of in mullitization is much higher than the glass volume at the the pellets, previously fired at 1 100 C and broken into.0 mm eutectic temperature. Therefore, the mullite transformation tem- diameter pieces, to determine the mullitization temperature as well perature is highly sensitive to the appearance of the glass phase in as any other phase transformation that could occur. The preferen- the system and must be a good expression of the eutectic nment,as well as the phase identification of the whiskers. temperature. The recent findings of Kolitsch, concerning the fined by analyzing the X-ray diffraction(XRD) pattern eutectic temperature of the LaAs system, show that higher RU200B, Rigaku, Tokyo, Japan) of the untreated pellet temperatures were obtained from old measurements surface at room temperature. The method used to measure the In addition to the mullite transformation peak at 1213C, the whiskers'c-axis coefficient of preferential orientation (CPO) erbia-doped sample shows several other endothermic peaks that evealed in the XRD pattern, calculates the preferential orientation are attributed to the formation of rare-earth silicate phases, among them Er, Si,O,, which was identified by its XRD pattern. There- coefficient"(MC), as described elsewhere. ' The microstructures of fore, the addition of lanthana, neodymia, erbia, and yuria de- the untreated surface of the pellets were observed by scanning creases the mullitization temperature. Although most rare earths electron microscopy(SEM)(Model DMX 980, Zeiss, Karlsruhe have been found to decrease the mullitization temperature and Germany) and the phases analyzed by EDS(Model QX 2000, Li induce whisker growth, the reported phase diagram of the LaAS Analytical Systems, High Wycombe, Buckinghamshire. U. K) system" and the absence of any other transformation temperature shown by DTA analysis were the reasons we chose the lanthana system for this study. Thus, the XRD pattern was expected to be Il. Results and Discussion simpler than that of erbia and, therefore, to allow for an easier quantitative analysis of the preferential alignment of whiskers. The The results of the present work were strongly influenced by the XRD pattern of the pellets fired at 1100oC confirmed the absence silica-rich glass melt that permeated the mullite body ar over the ceramic grains of the surface. This melt decre of mullite, while lanthania-doped pellets fired for 2.0 h in temperature of mullite formation, changing the mullite he1250-1700 C temperature range revealed that mullite was the tion and, at higher temperatures, the whiskers nucleation only main crystalline compound The presence of the glass phase responsible for the lowering of ate, and surface alignment. the mullitization temperature is well illustrated in the microstruc- There are significant differences in the few published articles eutectic compositions; however, the concentration of alumina does ures of the untreated surface of the pellet fired at 1250C and not exceed 24 mol%, whereas that of silica is less than 66 mol%, The phase diagram of the LnAS system, 8-10 for temperatures above the eutectic and in the concentration range of the present 1217C work has not been studied in sufficient detail to take into account mullite with both the Al, O / SiO,= 1.3 and the Al,O /Si0,=1.5 compositions, Considering the YAS system, about which more Exo details are available, o the phase diagram becomes more complex 1213°C as the temperature increases, even without considering the two mullite compositions. Moreover, the diphasic gel preparation route used in the present work does not promote the uniform dispersion of the components, contributing to the lack of thermodynamic equilibrium of the system under study, at least at the lower 122°C temperatures considered in this work. However, the high green density of the ceramic bodies, produced by 200 MPa isostatic pressing, allowed the grain-boundary phase to percolate among the mullite grains during firing. at least at higher temperatures, leading At temperatures close to the eutectic one, the grain-boune phase composition is expected not to be uniform because the 3.0 100011001200130014001500 boundary. Therefore, the results of the published phase diagran Temperature [C are not strictly applicabl atures close to the eutectic one. Most of the ceramic body, ever Fig. 1. Differential thermal analysis of pure Al,O/SiO,= 1.5 powders, after long firing times, is composed of stable Al,O/SiO2=I .5 as well as lanthana, neodymia, erbia, and yttria-doped Al, O/SiO mullite grains in every temperature range of the present study
Journal of the American Ceramic Society-Regiani et al. Vol. 85, No. I 王 E品 a0101601016170 Temperature rc] Fig 3. Temperature dependence of the coefficient of preferential orien ation(CPO)of whiskers on the pellet surface 0 000X magnification and appeared to be free of precipitates.At temperatures <1400 C, the AL,O/SiO, molar ratio of all of the mullite grains is very close to 1. 5. Above 1400 C, the microstruc tures also show an increasing number of alumina grains and B Our analysis of the grain boundary and triple points pha shows the SiO, and La,O, concentrations increasing with the firing temperature(see Table L, columns 2-4). The composition of the grain-boundary phase is assumed to result from the mixing of a glass phase and mullite of the Al,O /SiO,= 1.5 composition, and the composition of the glass phase considered here is close to the eutectic composition, La, O3Al-O3 6SiO,. The results of the ain-boundary analysis are, therefore, expressed as AL,O, SiO, mullite dissolved in the (LayO3)(Al,O3)-(SiO,)o go glass phase. The results of this calculation are shown in columns 6 and 7 of Table L. show owing the grain-boundary composition displacing toward the eutectic, with mullite decreases as the temperature increases. The only available explanation for the alumina and the anisotropic mullite grains with Al,O /SiO2= 1.3 composition found in the observed microstructures is the change in composition of the grain-boundary glassy phase as the temperature increases. The evolution of the grain-boundary composition induced by temperature can, thus, be represented as follows: GB glassy phase- La-richer glassy phase +(A2O/SiO2=1.5A2OSiO2=1.5) Fig. 2. SEM micrographs of the untreated surface of lanthania-doped pellets fired (a) at 1250C for 6.0 h and (b) at 1450C for 72.0 h. →1.3 mullite+ alumin Although the number of elongated Al,O/SiO,=1.3 mullite and alumina grains increased above 1400 C, their concentration in 1450C(Fig. 2), which shows mullite grains covered by an the sar ple was insufficient for detection by XRD. Abov amorphous phase. Even samples fired for a very long time(72 h) temperature, the well-defined 1.5 and 1.3 mullite compositions ive a similar microstructure (Fig. 2(b)), and also show the were in equilibrium, a result not found in the published phase absence of whiskers at 1450oC. To perform the EDS analysis of the grain boundary glassy phase at triple points and mullite grains. The anisotropic mullite grains with the Al,O, /SiO,=1.3 Columns 2-5 of Table I give the results of the analysis at -100 large aspect ratio, contrary to the case of whiskers. The c-axis um below the original surface. Along the course of such measure- orientation of the mullite grains in the temperature interval under ments, the grain boundary and triple points were observed under consideration is expressed by a small CPO(Fig. 3). According to Table L. Temperature Dependence of the Mullite Grain and the Grain-Boundary Glass Melt Composition Composition (molse Mullite content in grai Temperature (C) boundary (mole) LaOs come(mol%) 1350 49.0 148-1.32↑ 44.0 50.0 148-1.32 1600° 32.2 1.31(whisker) 23 Average of five analyses for each temperature, 'Anisotropic grains. Untreated surface
ary 2002 Nucleation and Growth of Mullite Whiskers from Lanthanum-Doped Aluminosilicate Melts the definition of CPO, a true random distribution of the c-axis The extermal surface of a pellet fired at 1600"C for 3 h wa orientation of the mullite grains would result in a zero value for the attacked by acid in two steps In step l, a short acid attack(Fig CPO. The roughness of the untreated pellet surface also contrib- 4(b) was made on the glass around the whiskers, leaving only the CPO in this temperature interval( Fig 3)is assumed to result Step 2 consisted of a deeper attack to remove the whiskers and the from a small deviation of the random c-axis distribution of mullite superficial layer of glass to expose the mullite grains of the grains on the pellets surface. substrate(Fig 4(c). The substrate's microstructure showed nor mal mullite grains of the Al, O,/SiO,= 1.5 composition and a few (2) Firing above 1500C anisotropIc grains. Whisker growth starts at 1500C(Figs. 3 and 4(a)), at which The microstructure shown in Fig 4(a)suggests a higher degree point the CPO rises steeply, reaching a maximum at 1650C. The of whisker alignment than that expressed by the CPO(Fig3).A observed preferential surface alignment, high aspect ratio, and similar behavior was also observed in the ErAS system. The constant thickness of almost 2.0 um are evidence that the explanation for this finding and also for the CPO increase with elongated grains observed on the surface are mullite whiskers. In temperature is pellet surface roughness, which is clearly shown in Fig. 5. The XRD data represents a very large area of the characterized by a high silica molar concentration close to the microstructure and expresses the average CPO over a rough eutectic one(see Table D) with a much lower viscosity that favors surface. Increasing the temperature causes the glass phase to the diffusion and growth of crystals spread more evenly over the entire surface and decreases the 3um Fig. 4. SEM micrographs of the untreated surface of lanthania-doped pellets(a) fired at 1500"C for 2.0 h; (b)after HF attack sufficient to remove the glass phase between the whiskers ("e denotes alumina grains); and (c) further acid attack, long enough to remove the whiskers and the glass layer, showing the mullite grains of the substrate. White grains are lanthania-rich
236 Journal of the American Ceramic Society-Regiani et aL. Vol. 85. No. I Slums Fig. 5. SEM micrograph of the untreated surface of 3.0-mol%o-lanthania- doped pellet fired at 1600 C for 30 min. The c-axis orientation of the ce roughness. surface irregularities. This also explains why the whiskers at 1500 C show a low CPO. The maximum cpo value at is associated with the simultaneous growth of platelets whiskers at 1700 C. These platelets are AlO/SiO= 1.5 and no longer whiskers and, therefore, do not contribute to the CPO. An identical behavior was identified in erbia-doped samples (3) Nucleation and Growth The firing of sufficiently dense ceramic bodies produces thin grain boundaries, allowing for a small volume of glass to percolate ng the ceramic grains. At higher temperatures, the concentra tion of the grain-boundary phase becomes progressively more reads over the pellet ' s ex Thick glass layers on the substrate surface would also be uring sintering with a large excess of liquid phase however, in this case, the pellets would lose their shape, which was not observed in the pellets fired at the highest temperature. The need for a glass-layer/air interface to grow mullite whiskers is an previous research work. 1.2 This behavior is independent of the pellets(a) after firing at 1600"C for 3 h("X"represents nucleation ceni experimental fact that has been proven by the results of this and Fig. 6. SEM micrographs of the untreated surface thania-doped nature of the rare earth and (b)after an acid attack sufficiently strong to remove the whisker layer. To better understand the nucleation and growth of mullite The rosace pattern was found all over the pellets surface investigation was made of the substrate layer after the growth of on the whisker growth substrate, with the center of each rosace mullite whiskers. We have already shown that alumina grains and corresponding to a whisker nucleation site. Because of a greater thickness than that of the glass layer. whiskers grow partially the alumina concentration which is smaller in the whiskers than in immersed in the glass phase, their growth aided by the surface the mullite Al,O/SiO,= 1.5 grains of the substrate. To decrease tension and fed through their external extremity. In this process. the rosace was developed from the center. Because of impinging, lated in which the Al,O,/SiO, composition was reduced to 1.35. growth ends when whiskers meet each other and the substrate As expected, whiskers grew on this pellet surface, but this time becomes covered with an alumina-rich rosace layer. Glass dis- without alumina grains( Fig. 6(a). The whisker layer was carefully solves the mullite substrate and loses mullite during whisker removed by successive acid attacks to reveal the microstructure of growth to feed the growth. If the substrate is covered by alumina Fig. 6(b). As a result of the acid treatment, the exact point-to-point rich rosace. its ability to maintain the 3: 2 mullite dissolution correlation between the two microstructures was lost: however the process decreases, as does the whisker growth process. Another aspect of the whisker growth process that can be whisker nucleation sites, as indicated by the "X"in Fig. 6(a). The earned from Fig. 6 is that a thin liquid layer between the whiskers EDS analysis of the white triangle-shaped parts of the rosace(Fi and the substrate is a necessary condition for the memory of the 6(b), showed an Al,O,/SiO, molar ratio of -3.0. We believe th rowth process to be well recorded on the substrate. When the xcess alumina of the rosace must have resulted from the OSiO, ratio exceeds 1.35, the alumina grains grow on the ition [substrate mullite grains (AL,O,/SiO,= 1.5))- pellet surface mixed with the mullite whiskers, and rosace are no ipitation I whiskers(Al,( O2=1.3)l process of whisker bserved. to define whether the behavior of whisk The microstructures su that the rosace shape reflects growth depends on the Al,O /SiO, ratio, experiments were made