Crystallization kinetics of La2O3–Al2O3–B2O3 glass–ceramic composites

(1)

Crystallization kinetics of La

2

O

3

–Al

2

O

3

–B

2

O

3

glass–ceramic composites

Chih-Lung Chen

a

, Wen-Cheng J. Wei

a,∗

, Andreas Roosen

b

a_{Institute of Materials Science & Engineering, National Taiwan University, 1 Roosevelt Road, Sect. 4, Taipei 106, Taiwan, ROC} b_{Institute of Materials Science, Glass and Ceramics, University of Erlangen-Nuremberg, Erlangen, Germany}

Received 11 April 2005; received in revised form 8 August 2005; accepted 14 August 2005 Available online 21 October 2005

Abstract

One glass formulation (L2 glass) with the composition of La2O3, Al2O3and B2O3in a molar ratio of 10:10:80 was selected to cofire with Al2O3

filler. The composites underwent a two-stage crystalline evolution in the temperature range of 800 to 975◦C. The crystallization kinetics of LaBO3

grains and the transformation to LaAl2B3O9 phase were investigated by DTA, XRD, SEM/EDS, and TEM. The results showed that the Al2O3

filler plays an important role as the heterogeneous sites of LaBO3 nuclei, and as reactant for the formation of flaky LaAl2B3O9 crystals. The

apparent activation energy of LaBO3-phase formation in L2 glass was 534 kJ/mol and reduced to 466 kJ/mol by the addition of Al2O3. The detail

Keywords: LAB glass; Composite; Microstructure; Transformation; Crystallization; Glass ceramics

1. Introduction

A low-firing temperature (<900◦C), lead-free, nonalkali, and low dielectric constant and low-dielectric loss system, La2O3–Al2O3–B2O3(LAB) glass–ceramic materials have been synthesized and investigated.1,2The composition was simplified from DuPont 943 tape system,3and developed for the applica-tion of low-temperature cofiring ceramics (LTCC). The process-ing window of the LAB glass, determined by various thermal and wetting measurements, indicated that the initial shrinkage temperature of the glass could be as low as 700◦C. In particular, one glass composition (L2 glass) with a composition of La2O3 10 mol%, Al2O3 10 mol%, and B2O380 mol% was identified to have the most suitable sintering properties when doped with Al2O3particles as a filler phase in a mass ratio of 30/70. The sin-tered composite has three major crystal phases, namely Al2O3, LaBO3crystals, and flaky LaAl2B3O9crystal after sintering at 850◦C.2

In addition to sintering, the crystallization of glass–ceramics during heat treatments is also interesting. Several papers4,5have reported the influence of seeding and reduction of the activation energy in␥- or ␪- to ␣-Al2O3transformation. The other cases,

∗_{Corresponding author. Fax: +886 2 2363 2684.}

E-mail address: [email protected] (W.-C.J. Wei).

e.g. ZrO2 nanopowder6, were used as nucleating agents for CaO–P2O5–SiO2glass system, of which the activation energy of apatitle/wollastonite/leucite phases could be decreased from 477 to 375 kJ/mol owing to the increase of nucleation sites. Beside the seeding effect, detail microstructural quantifications were conducted in order to investigate the crystallization mechanisms of nucleation and growth in many oxide systems.4–8

The aims of previous investigations1,2 _{were to study the}

processing characteristics of LAB/Al2O3glass–ceramic com-posites and to evaluate its potential for LTCC application. The aims of this study were to quantify the microstructural character-istics and crystallization kinetics of two crystalline phases of the glass–ceramic composites, so as to understand the crystallization behavior in the temperature region between 800 and 950◦C.

2. Experimental procedure

2.1. Materials and sample preparation

Two precursor powders were used. One was a La–Al–B–O glass (L2) powder in a composition shown inTable 1prepared by the steps specified in our previous work.1,2The other was an Al2O3powder (A32, Sumitomo Chemical Corp., Tokyo, Japan). The powder mixtures (e.g. L30A with 30 mass% Al2O3) were prepared by the following steps.

(2)

3326 C.-L. Chen et al. / Journal of the European Ceramic Society 26 (2006) 3325–3334 Table 1

Composition of L2 glass and alumina powders

L2 glassa _A32b La2O3 10 (mol%) – Al2O3 12 99.5+ Ba2O3 78 – SiO2 – 0.07 Fe2O3 – 0.04 Na2O – 0.04

Phase Amorphous ␣-Phase

Mean particle size (␮m) 1.56 1.04

a_{Analyzed by ICP-OES.} b _{Data given by manufacturer.}

The powders were firstly dispersed in the mixture of methyl ethyl ketone (MEK) and ethanol. The slurry was ball-milled for 24 h, then dried in an oven at 70◦C overnight. The admixture was sieved through 140 mesh and granulation before die-pressing. The green disk in the diameter of 5.0 mm and a thickness of ca. 7 mm was prepared by die-pressing at 80 MPa.

The L2 glass and L30A specimens were sintered at 800◦C for 1 h, and/or heat-treated at 850–950◦C for 10–30 min. The heating rate was normally set at 10◦C/min. The composites was sintered in air and cooled in air furnace.

The determination of crystallization kinetic parameters of L30A glass–ceramic composite was performed by nonisother-mal methods. The investigations of the DTA data with dif-ferent heating rates were conducted using pulverized L2 glass and L30A glass–ceramic samples. Although two crys-talline phases, equiaxial LaBO3 and flaky LaAl2B3O9, were sequentially formed during sintering, we only investigated the crystallization kinetics of the LaBO3phase. The kinetic behav-ior of the LaAl2B3O9 phase will be reported in the next paper.

2.2. Property characterization

2.2.1. Differential thermal analysis (DTA)

The DTA (Thermal analysis 2000 series, DuPont, USA) in NTU and another DTA measurement (Model DTA 910, DuPont Corp., USA) offered by the University of Erlangen-Nurmberg were used to investigate the transformation kinetics of the com-posites during continuous heat treatment. Glass transition tem-peratures (Tg) and crystallization peak temperatures (Tp) were determined using a heat rate of 10◦C/min.

2.2.2. Phase quantiﬁcation

The phase identification and quantification of the sintered composites were investigated by an X-ray diffractometer (XRD, Philips PW1830, Philips Instrument, Netherlands) using Cu K␣ radiation. The applied voltage and current were 30 kV and 20 mA, respectively. The scan speed was 3◦/min and each step was 0.04◦. The pure LaBO3 crystalline phase for quantifica-tion XRD analysis was prepared and verified with the following steps: A DTA test of the pure LaBO3 powder run at 850◦C was carried out. As no thermal effect was recorded until 2 h, and no difference in the X-ray patterns of crystallized standards

was found, a full crystallization in pure LaBO3 standard was assumed. The patterns of pure LaBO3powder and the mixture with calcined Al2O3were used for quantitative analysis. Most of LaBO3and␣-Al2O3peaks overlap between 30◦and 50◦, except the indexed peak (1 2 2) of LaBO3phase that could be chosen for quantification. The relative amount of crystallized phase pre-sented in the glasses was determined by comparing the major peak intensity ((1 2 2), 2␪ = 44.438◦_{) of the unknown specimen} with those of the standard specimens.

The crystallization kinetics and mechanism have been ana-lyzed in this study according to the Avrami analysis and modified methods. The values of crystallization dimensionality and acti-vation energy of LaBO3phase formation were also evaluated.

2.2.3. Microstructural characterization

The sample for SEM observation was ground and lapped with various diamond and Al2O3 suspensions. After polish-ing, the surface was cleaned and dried in a vacuum oven. Thin foils for TEM observation were prepared by typical ion-milling procedures by an ion miller (PIPS, Gatan Co., USA) until a thin, electron-transparent area was obtained. Scanning and transmission electron microscopes (field-emission SEM, Leo Instrument 1530, England; TEM, 100CXII, JEOL Co., Japan and HF-2000 FE-TEM, Hitachi Corp., Japan) were used. Besides, an analytical electron microscope (Tecnai 300 FE-TEM, Philips Instrument Corp., Netherlands) equipped with EDS was used to investigate the elemental distribution in each phase.

3. Results

Previous reports1,2 have identified two crystalline phases, LaBO3and LaAl2B3O9,awhich may grow in the L2 glass at the temperatures greater than 800◦C. The data of thermal analysis of the L2 glass and two LAB/Al2O3glass–ceramic composites are shown inFig. 1. The DTA curve of the pure LAB glass shows a distinct endothermic peak (Tg), representing a glass transition, and two exothermic peaks (TP1and TP2), representing the crys-tallization of the phases, but only one exothermic peak is found in L30A and L40A (L2 glass mixed with 40% Al2O3), respec-tively.

The endothermic glass transition reaction in the DTA curves was due to a change in the heat capacity attributed to the trans-formation of glass structures. The first exothermic peak depicted the onset crystallization temperature (To) and the crystallization peak temperature (TP1) of the LaBO3phase. It was noted that pure LAB glass had an additional exothermic peak, which should be the crystallization of the LaAl2B3O9phase. Besides, the DTA curve of L40A showed Tg, To, and Tp at higher temperatures (Tg= 760◦C, To= 848◦C, and TP1= 877◦C). Most significantly, the second exothermic peak became broad and faded away in the L30A and L40A curves.

a _{The phase is indexed as LaAl}

2O3(B4O10)O0.54, (JCPD file 87-0484), and

also indexed by JCPD file 47-1824 as (Ce,La)Al2B3O9. For easy presentation,

(3)

Fig. 1. DTA curves of pure L2 glass, L30A, and L40A glass–ceramic composites at a heating rate of 10◦C/min. The dotted line was taken as the base line of the curve.

Typical SEM micrographs of the L30A by different sintering conditions are shown inFig. 2. Homogeneous microstructure without porosity (measured density = 3.57 g/cm3) was observed inFig. 2(a), in which the Al2O3particles were in a darker contrast and well distributed in the matrix. The sintering temperature 800◦C was able to densify the matrix by a viscous flow of the La–Si–borate glass, but the refractory Al2O3particles did not change apparently.

The morphologies of the polished L30A obtained from sec-ond heat-treatment stage (850◦C/1 h) were also imaged by a back-scattering electron (BSE) mode, as shown in Fig. 2(b). The Al2O3particles became rounded at the edges (Fig. 2(b)) after treatment at 850◦C. This phenomenon implied that Al2O3 grain might dissolve into the glassy matrix during the second heat treatment. Therefore, small Al2O3 particles or the sharp corners of the Al2O3particles would dissolve. Besides, bulky LaBO3crystals grew more in volume when sintering was con-ducted at temperatures≥800◦C.

The matrix after treatment at 850◦C showed additional LaAl2B3O9 flaky crystals (in gray contrast) as shown in Fig. 2(b). The flaky crystals with the length longer than 1␮m grew with the orientations in specific pattern and distributed between the granular LaBO3and Al2O3phases. The details will be examined by TEM and reported in the following sections. 3.1. Crystallization kinetics of LaBO3phase

The LaBO3phase was the primary crystalline phase grown in the composite after sintering at ≤850◦C. The correspond-ing bright-field (BF) TEM images and diffraction pattern (DP) of the LaBO3 phase sintering at 800◦C for 30 min in air are shown inFig. 3. The equiaxial LaBO3crystals in the sizes of 0.15–0.45␮m showed distinct features from the matrix, and mostly surrounded by a glassy layer (Fig. 3(a) and (b)). The glassy layer had a distinct interface with the LaBO3grain,

imply-Fig. 2. SEM micrographs by back-scattered electron (BSE) imaging showing the component distribution and crystal phases of L30A sintered at (a) 800◦C/1 h and (b) 800◦C/1 h–850◦C/10 min in air at a heating rate of 10◦C/min. ing good wetting properties of the borate glass and good thermal expansion matching with the LaBO3grain. The high-resolution micrograph (Fig. 3(b)) illustrated the lattice image of (0 2 0) plane with∼0.7 nm spacing. As the DP shown in the insert, a broad diffraction ring was resolved, which should be contributed by the amorphous glass in the neighborhood of the LaBO3grain. Homogeneous and heterogeneous nucleations of the LaBO3 phase might occur concurrently at temperatures ≥800◦C. It was noted that the formation of LaBO3 phase started on the glass/Al2O3interface, evidently judging from the SEM results (Fig. 2(a)). The heterogeneous nucleation is the dominant mech-anism during sintering.

A quantification data of the crystalline LaBO3 phase as a function of the heating rate is shown inFig. 4. Nearly none of LaBO3grew in the matrix at a rate of 25◦C/min to 825–850◦C, but grew to the fraction of 0.8–0.9 at a slower rate of 5◦C/min. In order to consider the effect of quick crystallization, the trans-formation kinetics was analyzed using the method of continuous heating reported by Kissinger.9,10

Two sets of the DTA tests for either pure L2 glass or L30A glass–ceramic composite are shown inFigs. 5 and 6. The data

(4)

3328 C.-L. Chen et al. / Journal of the European Ceramic Society 26 (2006) 3325–3334

Fig. 3. TEM images showing the morphologies of L30A by 800◦C/1 h and the LaBO3crystals grown from the matrix: (a) two crystals and (b) grain/glass

interface and the inserted showing the DP of the LaBO3grain.

of the L2 glass showed distinguishable second exothermic peak in the DTA curves compared to those of L30A composite. The DTA results also showed that the crystallization of LAB started to form at a lower temperature than the case in the pure glass, possibly because of the nucleation sites offered by Al2O3filler. The point of heterogeneous nucleation will be discussed in Sec-tion4.1.

The Avrami exponent, n, could be determined from non-isothermal DTA data using the following equation:10,11

d ln[− ln(1 − x)]

d ln α |r = −n (1)

where x is the volume fraction of a crystallized phase tested at a heating rate of α. As a consequence, the treatments of

Fig. 4. Amount of LaBO3phase growing from L2 glass or L30A glass–ceramic

composite at specified temperatures plotted as a function of the heating rate.

ln[−ln(1 − x)] versus ln α at various temperatures of pure LAB glass and L30A composite were conducted and the results are shown in Fig. 7. The exponent, n, was determined from the slopes with the values of 2.5–2.7, implying that the crys-tallization could be one with three-dimensional (3D) growth of LaBO3 crystals.12 The analysis data of the crystalliza-tion mechanism are similar in pure LAB glass and in L30A composite.

(5)

Fig. 6. DTA curves of L30A glass–ceramic composite testing at different heating rates.

The thermal activated process could also be evaluated from the nonisothermal measurements using the following form:10 lnα n T2 p = mQ RTp+ const (2)

where m is the dimensionality of the crystal growth and R the gas constant. The value of m equals to n when the crystallization at different heating rates occurs on a constant number of nuclei for nonisothermal mode. In this study, the DTA samples were pre-heated around the nucleation temperature (725◦C) for 1 h, and then underwent various heating-rate treatments. It was assured that the number of nuclei was not dependent on the heating rate for the following two evidences. First, the DTA curves of the pre-and after-heat-treatment had the approximate identical Toand Tp points. The second, based on the SEM micrographs, as one typ-ical case shown inFig. 2(a), the number of LaBO3crystals per area were in the range from 4.1× 106/cm2to 5.3× 106/cm2, and remained roughly unchanged with time. This result indicated no formation of new crystals and implied a zero nucleation rate ( ´N = 0) during the heat treatment.

The apparent activation energy, Q, was evaluated from the nonisothermal measurements using this modified Kissinger method. A plot of ln (α3/T2) versus 1/T is shown in Fig. 8. The best fitting lines of two sets of DTA data were obtained and plotted in the figure. The QL2 (glass) and QL30A (L30A composite) values calculated from the slopes were 534± 15 and 466± 15 kJ/mol, respectively. The crystallization processes of pure glass with Al2O3 particles appeared at lower

acti-Fig. 7. Analysis results illustrating the amount of phase transformation plot-ted against the heating rate of pure L2 glass heating at 850◦C, and L30A glass–ceramic composite heating at either 825 or 850◦C.

vation energy (Q 60 kJ/mol) than that of the pure glass (L2).

3.2. Formation of LaAl2B3O9

In addition to LaBaO3phase, LaAl2B3O9phase was identi-fied in the matrix (Fig. 2(b) andFig. 9). As the TEM micrograph appeared, four regions (A, B, C, and D) were examined by electron diffraction, revealing regions A and C to be␣-Al2O3, region B to be LaBO3phase, and region D (flaky features) to be

Fig. 8. Analysis results of pure LAB glass (QLAB) and L30A glass–ceramic

(6)

Fig. 9. TEM bright field (BF) image showing the grain morphologies and phase distribution of L30A glass–ceramic composite sintered at 800◦C/1 h, then 850◦C/10 min in air at a heating rate of 10◦C/min. The grains of A (and C), B, and D were identified as Al2O3, LaBO3, and LaAl2B3O9phases, respectively

(see Section3.2).

LaAl2B3O9phase. The Al2O3grains were equiaxial and exhib-ited a size distribution of submicrometer. Besides, Al2O3grains showed very few defects, such as dislocation or faults included in the grains after sintering.

Flaky LaAl2B3O9 crystals were studied and revealed in Figs. 10–12. The TEM images showed that the anisotropic flaky crystals were mostly formed after annealing at ≥850◦C. The crystals showed a high aspect ratio with a thickness of ca. 50 nm. Centered dark-field (CDF) imaging shown inFig. 10(b) illus-trated that the crystals might have a similar orientation by a diffraction intensity of (0 0 2), as shown in the selected-area DP inFig. 10(c). The bright regions in the CDF micrograph also showed that the crystalline orientation between the neighbor-ing LaAl2B3O9crystals were close, implying that the growth of those flaky crystals came from the same nucleus.

When the L30A sample was heat-treated at 950◦C, which was 100◦C greater than the normal conditions used in the pre-vious data (Fig. 10), a series of SEM and TEM micrographs of the L30A composite were taken. The images showed that the flaky LaAl2B3O9crystals had grown to the lengths of several micrometers (Fig. 11(b)), but LaBO3grains were hardly found in the matrix. The LaBO3 grains must have reacted with the neighboring species to form flaky grains.

The results of a shrinkage test of the L30A composite were reported inFig. 11(c), of which the rate decreased at≥810◦C and turned to expand at temperatures >880◦C. The change in the shrinkage rate is due to the crystal growth of LaAl2B3O9. The formation of LaAl2B3O9crystals was contributed by a reaction of LaBO3with Al2O3filler and/or glassy matrix. The reaction took place at the temperatures≥810◦C, and became more evi-dent above 880◦C.

Detailed interfacial investigation of Al2O3/LaAl2B3O9/ Al2O3was conducted and shown inFig. 12. According to the compositional analysis, the Al and La elemental line scanning in Fig. 12(b) revealed a diffusion layer of about 30 nm, showing a

Fig. 10. TEM morphologies of LaAl2B3O9crystals grown in L30A sintered at

800◦C/1 h–850◦C/10 min. (a) BF; (b) CDF; (c) with the corresponding DP and zone pattern of [−1,1,0].

gradient Al2O3content between LaAl2B3O9/Al2O3interfaces. The LaAl2B3O9 phase was predominantly nucleated on the Al2O3surface. Furthermore, the amorphous phase was always found at the grain triple junction between LaAl2B3O9and Al2O3

(7)

Fig. 11. (a) SEM, (b) TEM micrographs showing the flaky crystals of L30A, and (c) dilatometric curve showing an obvious volume expansion starting at 880◦C. grains (Fig. 13). These glass phases could offer the diffusion path for the formation of LaAl2B3O9.

The interface between Al2O3/LaAl2B3O9is continuous. The d(1 1 2)of LaAl2B3O9crystal is 2.068 nm, which is close to d(0 0 6) of␣-Al2O3phase, 2.165 nm. The difference of the lattice space between LaAl2B3O9/Al2O3is ≤5%. The interface misfits for

Fig. 12. (a) TEM micrograph of L30A sintering at 800◦C/1 h–850◦C/10 min and (b) qualitative compositional profiles of Al and La elements near the Al2O3/LaAl2B3O9/Al2O3interface.

one (1 1 2) lattice plane every 25–26 layers, implying that the interface of LaAl2B3O9/Al2O3(Fig. 13)) is still semicoherent. Also, as the matching layers become more than 26 (ca. 50 nm), one new LaAl2B3O9 nucleus grows at the triple junction of Al2O3/LaAl2B3O9/glass. As a result (Figs. 10–13), the crys-tals of LaAl2B3O9showed a platy feature in two-dimensional radial extension from Al2O3surfaces.

4. Discussion

4.1. Heterogeneous nucleation of LaBO3

In order to assure the heterogeneous nucleation of LaBO3 phase, which is induced at the interface of Al2O3/L2 glass, one

(8)

Fig. 13. (a) TEM image of LaAl2B3O9/Al2O3grains with the inserts of one DP

illustrating the crystalline orientation of the Al2O3and (b) EDS spectrum of a

glass composition.

sample consisting of 70% L2 glass and 30% ultra-fine␣-Al2O3 powder (AKP-50, 0.3␮m average size, surface area 5.02 m2/gm) was prepared. After sintering at 800◦C for 1 h, the same sintering condition as L30A (Fig. 2(a)), the polished surface (Fig. 14) showed significant difference in the nucleation density of LaBO3 crystals in the sample. More LaBO3crystals (a higher crystalline density) could be found on the Al2O3/glass interface. Crystalline clusters in the matrix were also observed. The number density of LaBO3crystals was 1.7× 107/cm2, which was about two times higher than the density shown inFig. 2. The finer Al2O3filler offered more nucleation sites due to a higher specific surface area of Al2O3used for L30A.

The formation of LaBO3in L2 glass has an activation energy of 534± 15 kJ/mol, which is higher than that in L30A, because most of the nucleation takes place on the heterogeneous sites. The difference in the activation energy of QLAB and QL30A is about 68 kJ/mol, which implies that the formation of LaBO3 crystals on the Al2O3/glass interfaces in the L30A is easier than the growth in pure L2 glass. The difference in the activation energy between LAB glass and L30A glass–ceramic composite is mainly due to heterogeneous nucleation. The site density for nucleation is proportional to the available Al2O3/glass interface area.

Fig. 14. SEM micrographs in showing the equiaxial LaBO3crystals in a L30A

formulation with fine␣-Al2O3filler (average size 0.3␮m) sintered at 800◦C

for 1 h in air at a heating rate of 10◦C/min. 4.2. Transformation kinetics of LaBO3

It has been shown by Ozawa that the validity of Eq.(2)could be extended under certain circumstances for nonisothermal con-ditions. Two simple models for the case of crystal growth from pre-existing nuclei were proposed.13One is the crystal growth from impurity or nucleation agent, where the number of nuclei is independent upon the thermal history of the material. The second model was about the case of random (homogeneous) nucleation clearly separating from its following crystal growth. So the num-ber of the nuclei was changing during the crystal growth. The L30A glass–ceramic system in this study, therefore, belongs to the first model.

In this study, Avrami exponent n (Fig. 7) was measured, and was nearly independent of the heating rate. Possible values of n for different crystallization mechanisms are 1–4. If the values of m and n are equal for the cases, 3D crystallization is the predominant mechanism. The shape of LaBO3grains shown in Figs. 2 and 3supported the evidences of 3D growth of the crystals in the L30A sample. Besides, the nucleation mechanism is the interface control as n equals to 3.8

4.3. Effect of Al2O3on phase evolution

Previous analysis showed that L2 glass had no apparent reac-tion with Al2O3filler below (710◦C). The L2 glass did not form a liquid phase until the temperature reached Tg(710◦C). Due to the softening of the L2 glass, the glass penetrated into the space between Al2O3 particles, leading to some degrees of dimen-sional shrinkage and grain dissolution. However, the composite still contained some visible pores and grew LaBO3 grains at the glass/Al2O3 interface. At the temperatures higher than To (820◦C), the sintering rate of the LAB glass decreased, in the meantime, the LaBO3crystals started to react with Al2O3 par-ticles at the interfaces to form flaky LaAl2B3O9, and showed a fairly fast growing rate when treated at 900◦C. The final microstructure of L30A composite consisted of residual Al2O3

(9)

Fig. 15. Crystallization paths of LaBO3and LaAl2B3O9phases in the ternary

La–Al–B–O system.

grains, newly grown LaAl2B3O9plate grains, and La–Al–B–O glass.

The diffusion of Al3+ species by the dissolution of Al2O3 filler is the dominating factor for the reactions at temperatures above 710◦C. The reaction involves three possible stages. The first stage is the dissolution of Al2O3between 710 and 800◦C, of which the glass in a composition 10La2O3–10Al2O3–80B2O3 (L2) increases Al–O content as shown by the V1trajectory. In addition, the process will simultaneously trigger the formation of LaBO3(V2trajectory). Therefore, the combination of Al2O3 dissolution and the crystallization reaction of LaBO3 can be shown as below.

L2glass+yAl2O3→ 2xLaBO3+ Lx (3)

The composition of the L2 glass moves along the V3(V1+V2)

trajectory as the composition of Lx (10− x) La2O3− (10 + y) Al2O3− (80 − x) B2O3. The LaBO3 grains formed finely divided crystalline phase next to the glass and Al2O3grains.

The second possibility of the formation of LaBO3 phase is that the glass readily decomposes (a peritectic reaction) in accompany with the formation of Lgand Al2O3.

10La2O3− 10Al2O3− 80B2O3→ xLaBO3+ Lg+ Al2O3 (4) where Lg glass is a B2O3-rich (>95%) glass. The reaction is similar to the crystallization along the boundary of two primary fields. But no evidence of new Al2O3grains was found. There-fore, the reaction of Eq.(3) is more likely dominating in the system.

The third stage of the reaction is the formation of LaAl2B3O9 phase, which could be triggered as LaBO3 phase is formed. Two possible reaction routes are considered. One is a ternary peritectoid reaction, as the reaction of P1 shown inFig. 15. LaBO3+ L_x→ LaAl2B3O9+ Lg (5)

The other possible reaction involves the reaction of LaBO3 and Al2O3, shown as below:

LaBO3+ Al2O3+ Lp→ LaAl2B3O9 (6) where Lpis the peritectic liquid induced in the ternary diagram. The evidence showed that the volume fraction of Al2O3grains decreased (Figs. 2(b) and 11(a)) through only 10 mol% Al2O3 in the glass formulation, but 44 vol%bin the L30A. Therefore, the DTA curve (Fig. 1) showed that second exothermic crys-tallization peak (TP2) faded away and broadened, implying the LaAl2B3O9phase was able to form at lower temperatures owing to the availability of Al2O3ingredient.

Nanobeam EDS analysis obtained without the boron window presented the composition of the glass phase, which consisted of La, Al, B, and O elements. The LaAl2B3O9is thermodynam-ically stable and coexists with Al2O3and La–Al–B–O glass but not with B–O glass (Lg). Therefore, the EDS result supports that the phase formation of LaAl2B3O9is possibly following Eq.(6).

5. Conclusion

Two crystalline phases, LaBO3and LaAl2B3O9, dominantly grow in La2O3–Al2O3–B2O3glass–ceramic composites at tem-peratures above 710◦C by nucleation and growth mechanism. LaBO3 crystals growing in equiaxial shape undergo the het-erogeneous nucleation at specific sites on Al2O3grain surface. The crystallization activation energy of LaBO3 phase in pure L2 glass and L30A glass–ceramic composite were evaluated as 534± 15 and 466 ± 15 kJ/mol, respectively. The difference in the activation energy (68 kJ/mol) was due to the heterogeneous nucleation on Al2O3 grain surface. Besides, the number den-sity of LaBO3crystals remained constant with the heating time, indicting a zero nucleation rate in the process. The analysis on the microstructural evolution reported the values of Avrami con-stant (n) in the range of 2.5 to 2.7, confirming that the formation of the LaBO3crystal was a three-dimensional bulk crystalliza-tion. The microstructural observations by SEM/TEM were in agreement with the data by kinetics analysis.

LaAl2B3O9flaky grew at temperatures≥850◦C. The crystals were generated after the formation of LaBO3phase and grew from the triple junction of the LaBO3, Al2O3, and glass phase. The crystal growth of LaAl2B3O9showed similar platy features in radial directions from the Al2O3surfaces by the reaction of La–Al–B–O glass/Al2O3/LaBO3.

Acknowledgments

The authors like to thank Institute of Glass and Ceramics at University of Erlangen-Nuremberg, Germany, for kindly pro-viding laboratory facility, and the financial supported by PPP (0910044476) in Germany and NSC (92-2911-I-002-011) in Taiwan.

b_{Due to the presence of glassy phase, we are not able to estimate the density}

and molecular weight of the phase. Therefore, a rough estimation of alumina molar content in L30A is reported.

(10)

References

1. Wei, W. C. J., Chen, C. L. and Roosen, A., La2O3/Al2O3/B2O3 based

glass–ceramics for LTCC application. Key Eng. Mater., 2005.

2. Chen, C. L., Wei, W. C. J. and Roosen, A., Wetting, densification and phase transformation of La2O3/A2O3/B2O3based glass–ceramics. J. Eur. Ceram.

Soc., in press. 3. 943 tape, DuPont.

4. Shelleman, R. A., Messing, G. L. and Kumagai, M., Alpha-alumina trans-formation in seeded boehmite gels. J. Non-Cryst. Solids, 1986, 82, 277– 285.

5. Kao, H. C. and Wei, W. J., Kinetics and microstructural evolution of het-erogeneous transformation of␪-alumina to ␣-alumina. J. Am. Ceram. Soc., 2000, 83(2), 362–368.

6. Yu, B., Liang, K. and Gu, S., Effect of ZrO2 on crystallization of

CaO–P2O5–SiO2glasses. Ceram. Int., 2002, 695–698 (28).

7. Marrota, A., Buri, A., Branda, F. and Saiello, S., Nucleation and crystal-lization of LiO2–2SiO2 glass—a DTA study. In Advances in Ceramics,

Nucleation and Crystallization in Glasses, Vol 4, ed. J. H. Simmons, D. R. Uhlmann and G. H. Beall. American Ceramic Society, Columbus, OH, 1982, pp. 146–152.

8. Brown, W. E., Dollimore, D. and Galwey, A. K., Comprehensive chemical kinetics. In Reactions in the Solid State, Vol 22, Chap. 3, ed. C. H. Bamford and C. F. H. Tipper. Elsevier, Oxford, UK; New York, 1980.

9. Kissinger, E., Reaction kinetics in differential thermal analysis. Anal. Chem., 1957, 29(11), 1702–1706.

10. Matusita, K. and Sakka, S., Kinetic study on crystallization of glass by differential thermal analysis criterion on application of Kissinger plot. J. Non-Cryst. Solids, 1980, 741–746, 38–39.

11. Yinnon, H. and Uhlmann, D. R., Application of thermo-analytical techniques to the study of crystallization kinetics in glass-forming liquids. Part I: theory. J. Non-Cryst. Solids, 1983, 54, 253–275.

12. Henderson, D. W., Experimental analysis of nonisothermal transformation involving nucleation and growth. J. Therm. Anal., 1979, 15(2), 325–331. 13. Ozawa, T., Nonisothermal kinetics of crystal growth from pre-existing