Densification retardation in the sintering of La2O3-doped barium titanate ceramic

Densification retardation in the sintering of La2O3-doped barium titanate ceramic

Materials Science and Engineering A323 (2002) 167– 176 www.elsevier.com/locate/msea Densification retardation in the sintering of La2O3-doped barium ...

438KB Sizes 0 Downloads 21 Views

Materials Science and Engineering A323 (2002) 167– 176 www.elsevier.com/locate/msea

Densification retardation in the sintering of La2O3-doped barium titanate ceramic Ming-Hong Lin 1, Hong-Yang Lu * Department of Materials Science and Engineering, National Sun Yat-Sen Uni6ersity, Kaohsiung 80424, Taiwan Received 28 February 2001; received in revised form 9 March 2001

Abstract La2O3-doped TiO2-excess BaTiO3 has been studied for its pressureless-sintering behaviour. The resultant microstructure was analyzed by using both scanning and transmission electron microscopy. Second-phases of polytitanates, notably Ba6Ti17O40 and La2Ti2O7 (La2O3·2TiO2) are formed by solid-state reaction upon heating to the sintering temperature. Grain growth inhibition occurs at a threshold donor concentration of :0.30 mol.% La2O3 in the solid-state when samples are pressureless-sintered at 1215°C. The second-phase of monoclinic-La2Ti2O7 of the continuous nature located along grain-boundaries has been identified in the laser-sintered samples at the doping level of 0.50 mol.% La2O3. Densification retardation found in the La2O3-doped compositions is attributed to the presence of La2Ti2O7 that has effectively blocked the diffusion path for neck formation in the initial stage of sintering. The role of La2O3 on the solid-state sintering of BaTiO3 is discussed in favour of the solid-solution defect model. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Sintering; Microstructure; Boundaries

1. Introduction TiO2-excess powders have almost always been used [1] to study the sintering of BaTiO3 ceramics. However, sintering of such compositions is influenced by the BaTiO3 –Ba6Ti17O40 eutectic melt at : 1332°C [2] and Al2O3 –SiO2 –TiO2 (A – S – T) at an even lower temperature of :1250°C [3]. Densification, assisted by liquidphase, easily overwhelms that in the solid-state, which then becomes the predominant sintering mechanism at temperatures above the eutectic point. Consequently, studies into the effect of solid-state additives must be confined to a lower temperature range of B1250°C [4,5]. Both the A- [6] and B-site donors acted similarly to inhibit grain growth when coarsening was effectively suppressed by the doping level of 0.30 – 0.50 at.% [7]. * Corresponding author. Tel.: + 886-7-5254052; fax: +886-75256030. E-mail address: [email protected] (H.-Y. Lu). 1 Present address: Department of Mechanical Engineering, National Kaohsiung University of Applied Sciences, Kaohsiung 80782, Taiwan.

The critical concentration has been termed [4,7] as the grain growth inhibition threshold (GGIT). Beyond which, grain growth in La2O3-doped BaTiO3 ceramics is effectively hindered [4,7], microstructure refined and semiconductivity sharply reduced [6]. The solid-solution defect model [8] has usually been adopted [1,5 –7,9] to qualitatively interpret the enhanced sintering kinetics and the inhibition of grain-growth in donor-doped BaTiO3, despite the liquid-phase presence [5,9]. It is also often argued that cation vacancy generated upon donor doping [10] enhances densification by increasing the volume diffusivity [5,6,11]. Desu et al. [12] investigated highly donor-doped BaTiO3 and proposed a model of donor (Nd3 + )-segregated grain boundaries. In fact, a model of positive-charged Ti-enriched grain boundary and negative-charged space-charge layer in the donor-doped compositions has been suggested [13] for both BaTiO3 and SrTiO3. Solute-segregation impedes grain-boundary mobility, although the impuritydrag mechanism [8] inhibiting grain growth in sintering has not been substantiated [14]. It is known [15] that kinetic results alone cannot be adopted to suggest a sintering mechanism without ambiguity. An iterative

0921-5093/02/$ - see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 1 - 5 0 9 3 ( 0 1 ) 0 1 3 5 6 - 9

168

M.-H. Lin, H.-Y. Lu / Materials Science and Engineering A323 (2002) 167–176

procedure of identifying the rate-controlling mechanism from hot-pressing kinetics was proposed by Brook et al. [15], where the importance of microstructure analysis was emphasized. It is required to ensure the absence of liquid phase [16] for a solid-state sintering mechanism. Furthermore, excluding any second-phase is also necessary before the solid-solution defect model [8] may be adopted to account for densification improved with donor-doping. Yet, it has often been overlooked [1,5– 7,9] in the past for BaTiO3 [5]. A thorough microstructure analysis [17] of La2O3doped BaTiO3 ceramics using transmission electron microscopy (TEM) has not been reported in the literature. We have investigated the pressureless-sintering of La2O3-doped TiO2-excess BaTiO3 powders using a conventional tube furnace and CO2-laser. Densification retardation by La2O3-doping of \ 0.30 mol.%, more pronounced with laser-sintering, was found from both sintering techniques. Second-phases of Ba6Ti17O40 located at triple grain junctions and La2Ti2O7 along grain boundaries were identified by TEM. The sintering behaviour with donor doping is reported and the cause of the densification retardation associated with secondphases suggested.

2. Experimental procedure Commercial TiO2-excess 0.997 nonstoichiometric BaTiO3 powder (Ticon® HPB) supplied by TAM Ceramics (now Ferro Co., Niagara Falls, NY) was used in

Fig. 1. Schematic illustration of the setup for sintering using CO2 laser.

this study. La3 + was added by firstly dissolving La(NO3)3·6H2O in deionized water. The initial BaTiO3 powder was blended and mixed with the appropriate amount of dopant using a magnetic stirrer. It was then mixed with 1 wt.% PVB (polyvinyl butyral) binder in absolute alcohol before milling in a plastic jar with nylon-coated steel balls for 2 h. The mixed slurry was oven-dried, deagglomerated by using agate mortar and pestle and then passed through : 74 mm mesh. An appropriate amount of powder was die-pressed to 10 or 5 mm in diameter (f) in a WC-inserted steel die by applying a uniaxial pressure of 100 MPa. Green discs were then sintered in a conventional tube furnace at the heating-rate of 5°C min − 1. A CO2 laser of u= 10.6 mm (PRC, FH-3000, Landing, NJ) was used for laser sintering. The raw beam of 16 mmf was focused by a ZnSe mirror before defocusing to : 26 mm in order to cover the sample surface of 5 mmf entirely. The setup including high-Al2O3 refractory brick, in which a recess is made to accommodate a Pt-crucible for improving temperature uniformity in the sintered disc, is illustrated schematically in Fig. 1. Sintering temperature was calibrated against a radiation pyrometer (Chino IR-AH, Tokyo, Japan) by which the profile of input power and temperature was established. It was followed for each firing and checked periodically. Green compacts of arbitrary thickness have been laser-sintered to determine experimentally the optimal sample thickness at which the temperature uniformity is obtained. Polished cross-section was prepared for microscopy when average grain size across the sample thickness examined. It was found that uniform microstructure extended to : 0.5 mm. Samples were therefore die-pressed to : 0.5 mm in thickness for CO2-laser sintering. Sintered density was determined by applying the Archimede’s technique, where distilled water was used as the immersion medium. Sintering kinetic curves are established by plotting the final sintered density for various time or temperature when constant-heating rate was adopted for sintering. The identification of crystalline phases was made by X-ray diffractometry (XRD, Siemens D5000, Karlsrule, Germany) using Cuka radiation operating at 30 kV/2 mA with Ni filter. Sintered samples were mechanically ground and polished with SiC grits successively before diamond lapping to 1 mm roughness for microstructure analysis. Both reflected light optical microscope and SEM (JEOL JEM6400, Tokyo, Japan) were used for the observation of polished sections. Grain-boundaries were delineated by thermal etching at 1200°C or chemical etching using 1% HF solution where appropriate. Thin foils for TEM were prepared by the standard procedures of slicing to : 200 mm in thickness using a diamond-embedded saw, ultrasonic cutting to 3 mmf discs, mechanical polishing to 1 mm roughness, dimple-

M.-H. Lin, H.-Y. Lu / Materials Science and Engineering A323 (2002) 167–176

169

3. Results

3.1. Sintering kinetics

Fig. 2. Final sintered density versus dopant concentration for La2O3doped TiO2-excess 0.997 BaTiO3 compositions furnace-sintered at 1215°C up to 150 h.

Fig. 3. Sintered density-temperature dependence by (a) pressurelesssintering and (b) laser-sintering for La2O3-doped TiO2-excess 0.997 BaTiO3 compositions.

grinding to :20 mm thickness and Ar+-ion beam thinning to electron transparency. Observations were performed in JEOL AEM3010 operating at 300 kV.

The pressureless-sintering kinetic curves for the La2O3-doped TiO2-excess BaTiO3 compositions sintered isothermally at 1215°C for up to 150 h are presented in Fig. 2. It is clear that densification has been improved and higher final density obtained by donor doping. Final density is also found to increase with the La2O3doping level. The sintered density gained by La2O3doping has been improved incrementally when the doping level increases from 0.15, 0.30 to 0.50 mol.%. Figs. 3a,b give the sintered density increase with temperature by sintering at the constant-heating-rate (CHR) of 5°C min − 1 and laser-sintering, respectively when added with La2O3. CHR sintering does not exhibit the increasing trend of sintered density with La2O3-doping as that shown in Fig. 2 for isothermal sintering. It is represented in Fig. 3a that donor-doping has in fact retarded rather than enhanced densification when samples were sintered by CHR to temperatures 5 1350°C. Furthermore, the decrease in density is more pronounced with higher doping levels of 0.80 mol.%. The reduction of sintered density in the later stage termed ‘de-sintering’ [18,19] is also observed in La2O3doped samples for the undoped TiO2-excess compositions. The doping level of 0.15 mol.% La2O3 has resulted in greater de-sintering starting at :1325°C, lower than 1425°C of 0.30 mol.% La2O3 (Fig. 3a). For doping levels \ 0.30 mol.%, however, density reduction has disappeared even for sintering up to 1490°C, far exceeding the 1332°C eutectic point [2]. Higher dopant levels have delayed [17] the de-sintering until after higher sintering temperature or probably stopped it all together. The trend has also been observed in Y2O3doped samples where the second-phase Y2Ti2O7 is detected by TEM [17]. The dependence of sintered density upon temperature by two sintering techniques was compared. De-sintering of Dzrel : 2% in samples doped with 0.15 mol.% La2O3 is still noticeable at : 1325°C (Fig. 3b). Sintering by CO2-laser has become distinctive when the doping level is increased to 0.30 and 0.50 mol.% La2O3. It is clear that densification has been severely retarded from dopant of 0.30 mol.% onwards when de-sintering [19] is not observed. Samples doped with 0.30 mol.% La2O3 achieved higher final density, since no de-sintering had occurred up to 1500°C (Fig. 3b). Sintering kinetic curves using CO2-laser are difficult to establish owing to the faster densification initially. However, two distinctive types of the sintering kinetic curves can still be appreciated from the La2O3-doped compositions, as shown in Fig. 4. The similarity for the undoped and 0.15 mol.% La2O3-doped compositions is demonstrated by their sintering kinetics at 1390°C up to 300 s. They

170

M.-H. Lin, H.-Y. Lu / Materials Science and Engineering A323 (2002) 167–176

are both characterized by de-sintering at zrel : 95%. Sintered density increases very sharply in the initial 30 s to zrel :95% before again decreasing by Dzrel : 1–2% as firing at 1390°C continues. For compositions doped with 0.30 and 0.50 mol.% La2O3, however, initial sintering is much slower, judged by the sintered density obtained after sintering for 30 s. The fact that sintered density only increases progressively is better evidenced by the 0.50 mol.% La2O3-doped samples. Densification is again impaired in the initial stage of sintering by high La2O3-doping, although the final density of :94%, almost identical to that of the undoped TiO2-excess composition, is eventually obtained after a longer sintering time of 400 s. The indication is that donor-dopant of higher concentrations has taken longer sintering time to become effective on densification.

3.2. Microstructure obser6ation and X-ray analysis SEM micrographs, shown in Fig. 5, are from La2O3doped samples pressureless-sintered at 1215°C/100 h by CHR of 5°C min − 1. The large plate-like abnormally grown grains [4,5,21] characteristic of undoped TiO2-excess samples sintered at low temperatures are observed in the 0.15 mol.% La2O3-added samples, although they have been much reduced in number (Fig. 5a). The significance is that plate-like grain growth often found in the undoped samples [4,5,21] is suppressed by La2O3doping of GGIT ]0.30 mol.% (comparing Fig. 5a with 5b). Homogeneous microstructure of refined grains of Gav B 1 mm is obtained (Fig. 5b). Therefore, grain growth inhibition has been demonstrated at 1215°C, below any possible liquid eutectics in the BaO–TiO2 [2] and Al2O3 –SiO2 – TiO2 [3] system. It is clear that La2O3doping leading to grain growth inhibition is operative in the solid-state.

Fig. 5. Microstructures of La2O3-doped TiO2-excess BaTiO3 compositions, (a) 0.15 mol.% and (b) 0.30 mol.% sintered at 1215°C for 100 h (SEM– SEI).

Fig. 4. Laser-sintering kinetics for La2O3-doped TiO2-excess 0.997 BaTiO3 compositions.

Polytitanate [20] second-phases are located primarily at triple junctions or four-grain corners in 0.15 mol.% La2O3-doped samples laser-sintered at 1390°C for 60 s (Fig. 6). Some were identified [17] to be the monoclinic Ba6Ti17O40 after post-sintering annealing at 1200°C/30 h, as expected from the solidification of the BaTiO3 – Ba6Ti17O40 eutectic [2]. For 0.30 mol.% La2O3-doped samples sintered at 1350°C for 12 h, Fig. 7a,b illustrate, respectively second-phase Ba6Ti17O40 faceted along (001) and the corresponding SADP. Other polytitanate second-phases cannot be indexed to Ba6Ti17O40 unequivocally from the diffraction patterns, since the oc-

M.-H. Lin, H.-Y. Lu / Materials Science and Engineering A323 (2002) 167–176

171

currence [17] of the superlattice reflections and forbidden spots. In fact, solid-state-reacted polytitanates do not exhibit the characteristic lamellar structure of the eutectic solidification [5,17]. Note that the doping level is below the GGIT of :0.20 – 0.30 mol.% La2O3 [6,7,22] and these laser-sintered samples are quenched to room temperature in air. Serrated feature [5,17,23] along grain-boundaries (indicated in Fig. 6) represents the crystallographic microfaceting along {111} and {100} of BaTiO3 [5,17]. It also indicates that the eutectic liquid penetrating along grain-boundaries has solidified upon cooling to a mixture of polytitanates with residual siliceous glassy phase [5,17]. The microstructural feature rounded grain corners is also the characteristic of liquid-phase sintering. These samples, buff in color, are insulators having the room temperature resistivity of \108 V mm − 1. The high-temperature defect structure in the quenched samples is likely to be dominated by cation vacancies [24]. If La3 + dissolution into the BaTiO3 lattice was a kinetically controlled process [25] has been investigated by determining the lattice constants for samples doped with 0.80 mol.% La2O3. The (c/a)-ratio is still decreasing monotonically (Fig. 8a) with sintering temperature, although the c- and a-constants do not follow such a fashion. The indication is that 0.80 mol.% La2O3 is not completely dissolved in the solid-state even after heating to 1490°C by the heating rate of 5°C min − 1. Furthermore, the dissolution of La3 + into the BaTiO3 lattice has resulted in the depression of the c-axis, while the a-axis remains almost unaltered. Similar trend of the c- and a-axis is also observed for sintering at 1350°C for 12 h. The (c/a)-ratio decreases steadily with the increasing La2O3-level (Fig. 8b). The BaTiO3 lattice has also been modified towards pseudo-cubic (of c/a B 1.005). In fact, kinetically controlled dissolution in Fig. 7. Second-phase Ba6Ti17O40 in 0.30 mol.% La2O3-doped samples sintered at 1350°C for 12 h (a) BF image and (b) the corresponding SADP (TEM).

Fig. 6. TEM BF images of 0.15 mol.% La2O3-doped TiO2-excess BaTiO3 compositions laser-sintered at 1390°C for 60 s, second-phase located at a four-grain corner and ledges along grain-boundaries.

the solid-state can also be inferred from the 8 mol.% La2O3-doped samples at 1300°C from Fig. 8(c). It appears that dwelling at 1300°C for \20 h is required to complete the dissolution of La2O3, although the solid solubility in BaTiO3 could not be determined from the present results. Ba2TiO4 is detected by XRD from La2O3-doped samples sintered at 1300°C for 20 h. It suggests that La3 + substituting for the A-site results in Ba2 + exsolution and the powder composition moves towards BaO-excess. The exsolved Ba2 + reacts with the initial BaTiO3 forming the hygroscopic Ba2TiO4 [5,9]. These samples, similar to the BaO-excess samples [5], are too friable for preparing TEM foils when retrieved

172

M.-H. Lin, H.-Y. Lu / Materials Science and Engineering A323 (2002) 167–176

from the furnace after overnight cooling. Consequently, it is not possible to locate the second-phase Ba2TiO4 particles under the microscope [5] nor to evaluate its effect on grain growth inhibition. Second-phase (indicated in Fig. 9a) other than the polytitanates has also been detected in 0.30 mol.% La2O3-doped samples sintered at 1350°C for 12 h. However, the grain boundary phase of : 200 nm in

Fig. 9. Continuous second-phase(s) at three different locations of (a) and (b) along the grain-boundaries of laser-sintered 0.50 mol.% La2O3-doped TiO2-excess BaTiO3 composition (TEM – BFI).

Fig. 8. Dependence of the lattice constants and (c/a)-ratio with (a) sintering temperature, (b) La2O3 dopant concentration at 1350°C for 12 h and (c) change of unit cell volume of 8.0 mol.% La2O3-doped TiO2-excess BaTiO3 upon annealing at 1300°C.

thickness between two tetragonal BaTiO3 grains exhibiting the characteristic ferroelectric domain contrast cannot be identified unequivocally by the conventional SADP. The existence of such a continuous grainboundary second-phase, as opposed to the discrete particles in the Y2O3-doped TiO2-excess samples [17], is better illustrated in three different locations of the same foil prepared from 0.50 mol.% La2O3-doping and lasersintered at 1500°C for 60 s. Continuous second-phase with no ferroelectric domains is clearly discernible in Fig. 9a. Rigorous tilting under the microscope is usually necessary to appreciate their existence, particularly for its minute amount. The tetragonal BaTiO3 grain, shown in Fig. 9b, also appears to contain secondphase(s) along its boundaries with other grains. Those

M.-H. Lin, H.-Y. Lu / Materials Science and Engineering A323 (2002) 167–176

contained in the triple-grain junction were investigated further with high-resolution TEM. Fig. 10a exhibits four distinctive regions of (A– D) in the vicinity of the triple-grain junction and along grain-boundary, as indicated. Lattice-fringe imaging of region (B) given in Fig. 10a reveals the interface of (A)– (B). Region (A) of d101( , d110 and the angle of 80.5° between the two planes is consistent with the monoclinic-La2Ti2O7 (La2O3·2TiO2). Regions of (B– D) and those encircling the BaTiO3 grain (Fig. 9b), although not identified under such imaging conditions, are probably of the same phase or of compositions slightly deviated from the stoichiometry. It appears that, not only the polytitanates, but also La2Ti2O7 have influenced the sintering of La2O3-doped TiO2-excess compositions.

173

4. Discussion Controlled grain growth is often attempted by using additives in the solid-state [8]. Its mechanisms can be categorized [26] into two distinctive regimes: (1) solid-solution and (2) second-phase. The (dz/dG)-ratio [8,26] may be modified favourably when the densification rate is enhanced or/and grain growth rate suppressed in sintering. Impurity segregation [13] of the solid-solution regime may act in a similar way to second-phase pinning when the moving grain boundaries are being dragged and their mobility reduced. From the defect concentration modified by donor-doping, sintering is affected by a different mechanism in the solid-solution regime. It promotes densification by improving the bulk diffusion [5,10]. How fast donor oxide is dissolved in the BaTiO3 lattice at sintering temperatures (Fig. 8c) then becomes determining for the densification enhancement to be effective. Whether the donor oxide reacts with BaTiO3 or other trace impurities forming other second-phases upon heating may also contribute to reduce the concentration of the extrinsic defect. Besides, the principal charge compensation mechanism is temperature-dependent [17,24], it alters from predomination of cation vacancy at low temperatures of B1220°C to a mixed scheme, then to electron compensation at \1500°C. It therefore appears that the two factors of (1) dissolution kinetics and (2) possible reactions occurring upon heating may have modified the role of donor-oxide La2O3 added purposefully to improve the sintering of BaTiO3.

4.1. Dissolution kinetics

Fig. 10. Grain-boundary second-phases (a) designated as A – D and (b) lattice-fringe imaging for region B in (a) (TEM).

It is known that the existence of liquid eutectics [3,5,8,9] and solute segregation [12,13] affects [26] the sintering kinetics decisively. Enhanced densification in the sintering of BaTiO3 ceramics was often discussed [4,5,11,22,23,25] on the basis of the solid-solution defect model [8,17] adopting the (dz/dG)-ratio [27] argument. The dissolution of La2O3 is, in fact, kinetically controlled, as indicated from the benefit of pre-firing [25] and clearly demonstrated by the decreasing of the (c/a)ratio progressively, as shown in Fig. 8a–c. It is also often taken as if the dopants were dissolved almost immediately since sintering, usually performed using a slow heating rate [6,22], has allowed the dissolution to complete upon heating to sintering temperatures. Refined microstructure (Fig. 5) is indeed obtained from sintering at 1215°C using constant heating rate in the La2O3doped compositions. The fact that grain growth inhibition occurs at sintering temperatures lower than the A–S– T eutectic of 1250°C [3] suggests that it has occurred by a solid-state mechanism. Isothermal sintering by CHR to and dwelling at 1215°C (Fig. 2) has indeed shown that increasing La2O3-doping level is

174

M.-H. Lin, H.-Y. Lu / Materials Science and Engineering A323 (2002) 167–176

beneficial, while the additive effect is reversed by using laser of very fast heating rate (Figs. 3 and 4). The effect is more pronounced in laser-sintering (Fig. 3b) and less so in sintering with CHR (Fig. 3a), since longer dwelling upon heating up to sintering temperatures has allowed better dissolution of La3 + , the formation of (Ba,La)TiO3 solid-solution and the generation of the extrinsic defects to occur. Barium vacancy has been suggested [5] to be the rate-determining species from the sintering of undoped BaTiO3 compositions. The dissolution of dopants into BaTiO3 forming (Ba,La)TiO3 solid-solution, therefore acts directly to improve densification by the barium vacancies generated for charge compensation. It is inferred that the extrinsic defects thus created by donor-doping have contributed favourably to sintering by increasing the densification rate (dz/dt) in the solid-state. Indeed, pre-firing of La3 + -doped compositions at 1300°C for 3 h reported [15] to have better sintering results was attributed to the enhanced solid-solubility of La3 + and cation lattice-diffusivity in donor-doped BaTiO3. Dissolution kinetics of La2O3 in BaTiO3, however, cannot explain the hindrance of densification (Fig. 3a,b and Fig. 4) at temperatures below the lowest eutectic point of 1332°C in the system BaO– TiO2 [2]. On the contrary, higher dopant levels resulted in lower sintered density (Fig. 3a). It follows from the solid-solution models that the extrinsic defects associated with the rate-determining mechanism of BaTiO3 sintering [5] have become less effective in assisting densification in the initial stage. Hampering of densification depending on temperature, heating-rate (Fig. 3a,b) and dwelling time (Fig. 4) is more pronounced in laser-sintering. Enhancement of densification in the solid-solution regime determined by the dissolution kinetics of La2O3 has in fact been deferred to a later stage of sintering (Fig. 3a,b and Fig. 4). De-sintering [19] was also delayed until 1400°C for 0.40 mol.% La2O3 (Fig. 3a) and did not occur at : 1500°C for \0.40 mol.% La2O3 (both are higher than the Ba6Ti17O40 –BaTiO3 eutectic point at 1332°C). The retarding effect by La2O3-doping is not at all significant for low donor levels of 0.15 mol.%, particularly when the eutectic liquid emerges at \1332°C. It appears that increasing the doping level to \0.30 mol.% (Fig. 3a) resulted in hampering densification may have been due to the formation of the second-phase La2Ti2O7 (Figs. 9 and 10).

4.2. Sca6enging of La2O3 for TiO2 As for the second-phase of La2Ti2O7 in sintered donor-doped BaTiO3 ceramic [6,22], in addition to Ba6Ti17O40 [17] and the residual glass [5], it has not been reported in the literature.

Solid-state reactions between the excess TiO2, La2O3 and BaTiO3 are expected to take place concurrently during the heating-up stage. It leads to the formation of second-phases of Ba6Ti17O40 [5,17] (and polytitanates) (Fig. 7) and La2Ti2O7 (Fig. 10). The donor oxide La2O3 and the initial BaTiO3 powder particles are scavenging for the excess TiO2 in the starting powder, which would have made the powder move towards stoichiometric composition progressively as heating continues. Both second-phases may act to hinder grain-boundary migration by pinning when the ceramic is sintered below the eutectic temperatures. Furthermore, La2Ti2O7 formation has also consumed the limited amount of La2O3 available from doping. Consequently, its formation also reduces the concentration of the extrinsic defects that could act to improve densification in the solid-state. Therefore, the formation of La2Ti2O7 mitigates the donor-doping effect in the solid-solution defect regime, particularly in the initial stage of sintering. Densification retardation in laser sintering by adding La2O3 to : 0.30 mol.% (Fig. 4) is likely to have been imposed by the La2Ti2O7 grains of continuous nature located along grain-boundaries (Fig. 9a,b). These second-phase grains have effectively blocked the diffusion path during the initial neck formation. The experimental evidence of La2Ti2O7 (Fig. 10b), densification retardation (Figs. 3 and 4) and grain growth inhibition (Fig. 5) clearly support the view that La2O3 donor-oxide acts to improve sintering in the solid-solution regime. Prior to its dissolution into the BaTiO3 lattice, however, La2O3 reacting with the excess TiO2 is detrimental to densification. The eutectic liquid of BaTiO3 –Ba6Ti17O40 composition is also reduced in quantity due to La2O3 scavenging for the excess TiO2 has prevented forming Ba6Ti17O40 upon heating. It is also indicated by the postponement of de-sintering whose occurrence coincides with the formation of the eutectic liquid (Fig. 3a,b).

4.3. Second-phase pinning by La2Ti2O7 Second-phase pinning [28] has been proposed to account for the grain growth inhibition in systems such as MgO-added Al2O3 if the MgAl2O4 particles have indeed been formed [29], although its existence is disputable. The effectiveness of second-phase pinning depends on the size and distribution of the immobile particles leading to the Zener–Smith limit in normal grain growth [26]. However, the localization of La2Ti2O7 and its continuous nature (Fig. 9a,b) suggest otherwise, particularly when La2Ti2O7 has not been detected in samples of lower doping levels and sintered by CHR. The refined microstructure obtained from La2O3-donordoping (Fig. 5b) could not have been achieved solely by second-phase pinning or other mechanisms associated with the second-phase, unless La2Ti2O7 acts to affect

M.-H. Lin, H.-Y. Lu / Materials Science and Engineering A323 (2002) 167–176

the surface energy anisotropy of BaTiO3. The growth mechanism of plate-like grains [4,5] owing to the {111} planes of low solid-to-vapor surface energy has been favoured [21]. In fact, the excessive TiO2 and the La2O3 additive would not be sufficiently homogeneous and ample in quantity to distribute completely over the powder surfaces (Fig. 10a) in order to exert effective pinning collectively. It is possible that second-phase pinning by La2Ti2O7, if it occurs at all, may have been a transient effect which diminishes eventually when the coalescence of these grains occurs upon heating up to higher temperatures or its re-dissolution upon sintering for longer duration. Indeed, annealing for 20 h is found necessary to achieve complete dissolution, as evidenced by the unit cell volume (Fig. 8c). On the other hand, densification enhancement by the solid-solution defect mechanism is likely to be operating throughout the entire sintering process. The effectiveness is, however, dependent upon the dissolution kinetics of La3 + to the BaTiO3 lattice. For pre-firing [25] to be beneficial, the formation of La2Ti2O7 by solid-state reaction would have to be suppressed, bearing in mind that pore size and distribution may also be determining [8], since the neck formation is obstructed by La2Ti2O7. The controlling factor is therefore the dissolution of La2O3 competing kinetically with the two solid-state reactions forming second-phases. Whether La2Ti2O7 increases the eutectic temperature related to the BaO– TiO2 system is not yet certain [17]. Although de-sintering [19] occurs at the eutectic temperature of :1332°C, (as indicated in Fig. 3a for 0.15 mol.% La2O3-doped) it coincides [17] with the occurrence of the polygonal grains [5,7,17] signifying liquidphase-assisted sintering clearly having been delayed to the higher temperature of : 1425°C for 0.30 mol.%. For doping levels exceeding 0.30 mol.%, de-sintering has not been observed up to :1500°C (as shown in Fig. 3a). The likelihood is that the eutectic composition and temperature has altered from BaTiO3 – Ba6Ti17O40 of :1332°C [3] in BaO– TiO2 system to BaTiO3 – La2Ti2O7 or BaTiO3 – Ba6Ti17O40 – La2Ti2O7 of \ 1500°C in BaO –TiO2 – La2O3 system. The formation of La2Ti2O7 as well as Ba6Ti17O40 would have consumed the limited amount of TiO2 available from the initial powder and moved the composition towards the stoichiometric BaTiO3. Indeed, the unidentified polytitanates [17] may have associated with the pseudo-ternary system of BaTiO3 – Ba6Ti17O40 – La2Ti2O7. It also explains why the solid-state-reacted Ba6Ti17O40, exhibiting no lamellar feature characteristic to that solidified from the eutectic melt [5], has survived the sintering at 1350°C for 12 h (Fig. 7a). The lowest eutectic temperature in the system containing La2Ti2O7, however, cannot be determined from the present results without detailed DTA analysis. Nonetheless, that the two second-phases of Ba6Ti17O40 and La2Ti2O7 are

175

competing for the excess TiO2 during heating up is certain. The actual process would have been the concurrency of both solid-state reactions owing to the chemical inhomogeneity of the excess TiO2, as well as the La2O3-dopant in the powder mixture. The densification enhancement by solid-solution mechanism is delayed until after the re-dissolution of La3 + takes place upon long annealing.

5. Conclusions Grain growth inhibition threshold occurs at :0.30 mol.% in La2O3-doped BaTiO3 ceramic. Solid-state reaction of the excess TiO2 with BaTiO3 forming polytitanates of predominantly Ba6Ti17O40 and that with La2O3 forming La2Ti2O7 have both occurred during the sintering of La2O3-doped TiO2-excess BaTiO3 compositions. The second-phase of monoclinic-La2Ti2O7 grains has been identified by TEM. They appear to be highly localized owing to the chemical inhomogeneity of TiO2 in the starting BaTiO3 powder. Densification retardation in the TiO2-excess BaTiO3 compositions doped with La2O3 of concentrations exceeding the GGIT is attributed to the slow dissolution kinetics of La3 + into the BaTiO3 lattice and the formation of second-phase La2Ti2O7.

Acknowledgements Funding support by the National Science Council of Taiwan through NSC-80-0405-E110-018, 82-0405E110-029 and 83-0405-E110-007 is acknowledged.

References [1] T.F. Lin, C.T. Hu, I.N. Lin, J. Am. Ceram. Soc. 73 (1990) 531. [2] K.W. Kirby, B.A. Wechsler, J. Am. Ceram. Soc. 74 (Suppl. 8) (1991) 841. [3] Y. Matsuo, H. Sasaki, J. Am. Ceram. Soc. 52 (1971) 471. [4] L.A. Xue, Y. Chen, E. Gilbart, R.J. Brook, J. Mater. Sci. 25 (1990) 1423. [5] M.H. Lin, J.F. Chou, H.Y. Lu, J. Eur. Ceram. Soc. 20 (2000) 517. [6] C.J. Peng, H.Y. Lu, J. Am. Ceram. Soc. 71 (1988) C44. [7] L.A. Xue, Ph.D. Thesis, University of Leeds, 1987. [8] R.J. Brook, Proc. Br. Ceram. Soc. 32 (1982) 7. [9] H. Mostaghaci, R.J. Brook, Br. Ceram. Trans. J. 84 (1985) 203. [10] G.V. Lewis, C.R.A. Catlow, R.E.W. Casselton, J. Am. Ceram. Soc. 68 (1985) 555. [11] T.F. Lin, C.T. Hu, I.N. Lin, J. Mater. Sci. 25 (1990) 3029. [12] S.B. Desu, D.A. Payne, J. Am. Ceram. Soc. 73 (1990) 3407. [13] Y.M. Chiang, T. Takagi, J. Am. Ceram. Soc. 73 (1990) 3278. [14] S.B. Desu, D.A. Payne, J. Am. Ceram. Soc. 73 (1990) 3391. [15] R.J. Brook, E. Gilbart, D. Hind, J.M. Vieira, in: S. Pejovniv, M.M. Ristic (Eds.), Sintering-Theory and Practice, Elsevier, Amsterdam, 1982, p. 585.

176

M.-H. Lin, H.-Y. Lu / Materials Science and Engineering A323 (2002) 167–176

[16] C.J. Ting, H.Y. Lu, Acta Mater. 47 (1999) 831. [17] M.H. Lin, PhD Thesis, National Sun Yat-Sen University, Taiwan, 1998. [18] C. Herard, A. Faivre, J. Lemaitre, J. Eur. Ceram. Soc. 15 (1995) 135. [19] M. Demartin, C. Herad, C. Carry, J. Lemaitre, J. Am. Ceram. Soc. 80 (1997) 1079. [20] E. Tillmanns, W.H. Baur, Acta Crystallogr. B26 (1970) 1645. [21] G. Kastner, R. Wagner, V. Hilarius, Philos. Mag. A69 (1994) 1051. [22] C.J. Ting, C.J. Peng, H.Y. Lu, S.T. Wu, J. Am. Ceram. Soc. 73 (1990) 329. [23] G.M. Dynna, Y.M. Chiang, Sintering of advanced ceramics, in: C.A. Handwerker, J.E. Blendell, W.A. Kaysser (Eds.), Ceramic

[24] [25] [26]

[27]

[28] [29]

Transactions, vol. 7, American Ceramic Society, Westerville, OH, 1990, p. 547. J. Daniels, K.H. Hardtl, D. Hennings, R. Wernicke, Philips Res. Rep. 31 (1976) 487. H.W. Hsieh, T.T. Fang, J. Mater. Sci. Lett. 10 (1991) 276. R.J. Brook, in: F.F.Y. Wang (Ed.), Ceramic Fabrication Processes, Treatise on Materials Science and Technology, vol. 9, Academic Press, New York, 1976, p. 331. C.A. Handwerker, R.M. Cannon, R.L. Coble, Structure and properties of MgO and Al2O3 ceramics, in: W.D. Kingery (Ed.), Advanced Ceramics, vol. 10, American Ceramic Society, Colombus, OH, 1984, p. 619. S.J. Bennison, M.P. Harmer, J. Am. Ceram. Soc. 73 (1990) 833. K.L. Garilov, S.J. Bennison, K.R. Mikeska, R. Levi-Setti, Acta Mater. 47 (1999) 4031.