Role of lithium borosilicate glass in the decomposition of MgTiO3-based dielectric ceramic during sintering

Article abstract of DOI:10.1016/j.materresbull.2005.10.020

The decomposition behaviour of 0.9MgTiO₃-0.1CaTiO₃ dielectric ceramic during a liquid phase sintering by lithium borosilicate (LBS) glass was studied. The decompositions of MgTiO₃ into MgTi₂O₅ and Mg<

Full text of DOI:10.1016/j.materresbull.2005.10.020

Materials Research Bulletin 41 (2006) 1206–1214
www.elsevier.com/locate/matresbu
Role of lithium borosilicate glass in the decomposition
of MgTiO3-based dielectric ceramic during sintering
a
a
Hee-Kyun Shin a, Hyunho Shin b, , Hyun Suk Jung , Seo-Yong Cho ,
*
Jeong-Ryeol Kim a, Kug Sun Hong a
a School of Materials Science & Engineering, Seoul National University, Seoul 151-744, Republic of Korea
b Department of Ceramic Engineering, Kangnung National University, Kangnung 210-702, Republic of Korea
Received 15 September 2005; accepted 25 October 2005
Available online 28 November 2005
Abstract
The decomposition behaviour of 0.9MgTiO3–0.1CaTiO3 dielectric ceramic during a liquid phase sintering by lithium
borosilicate (LBS) glass was studied. The decompositions of MgTiO3 into MgTi2O5 and Mg2TiO4 were apparent during the
sintering although the reactions are thermodynamically unfavourable in glass-free compositions. The role of the LBS glass in
favouring the decomposition reaction was investigated in terms of the thermodynamic activity of the reaction product in the glass.
The decomposition reactions were not necessarily harmful because of the high dielectric performance of the decomposition
products, MgTi2O5 (17.4 permittivity; 47,000 GHz quality factor) and Mg2TiO4 (14.4 and 55,000 GHz, respectively).
# 2005 Elsevier Ltd. All rights reserved.
Keywords: A. Ceramics; D. Dielectric properties
1. Introduction
MgTiO3–CaTiO3 is a well-known ceramic material in low temperature co-fired ceramic (LTCC) technology [1–3].
The addition of glass to the MgTiO3–CaTiO3 ceramic composition is a convenient means for lowering the sintering
temperature from about 1350 8C. As the glass addition degrades the dielectric properties as well, the development of a
suitable glass system for the low temperature sintering has to be carried under the constraint of minimizing the
degradation of the dielectric properties. Thus, it is prerequisite to understand the material reactions during the sintering
process under the presence of glass, because these reactions influence the densification process as well as the resultant
dielectric properties.
In the present study, the decomposition reaction of 0.9MgTiO3–0.1CaTiO3 (MCT) ceramic was investigated
during the sintering process under the presence of a new sintering agent glass system, lithium borosilicate (LBS)
glass. This composition is an extension to previously studied borosilicate glasses RO–B2O3–SiO2 (R = Zn, Ba)
[4–6]. This work will demonstrate that the decomposition of MgTiO3 is thermodynamically unfavourable, but it
can take place under the presence of LBS glass (2MgTiO3 = Mg2TiO4 + TiO2 and 2MgTiO3 = MgTi2O5 + MgO).
The role of glass in favouring the decomposition reactions has been investigated in terms of thermodynamic
* Corresponding author. Tel.: +82 33 640 2484; fax: +82 33 640 2244.
E-mail address: hshin@kangnung.ac.kr (H. Shin).
0025-5408/$ – see front matter # 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.materresbull.2005.10.020

H.-K. Shin et al. / Materials Research Bulletin 41 (2006) 1206–1214
1207
activities of MgO and TiO2 in glass. It is also reported that the decomposition reaction is not necessarily harmful
because of the high dielectric performance of the decomposition products MgTi2O5 and Mg2TiO4.
2. Experimental procedure
Stoichiometric amounts of MgO (High Purity Chemical Lab., Japan), TiO2 (rutile, High Purity Chemical Lab.), and
CaCO3 (High Purity Chemical Lab.) powders (all were 99.9% pure) were mixed, ball-milled, dried, and calcined at
1100 8C for 2 h to prepare either MgTiO3 or CaTiO3 compounds. Li2O, B2O3, and SiO2 powders (LBS) were mixed
with a weight percent ratio of 3:6:1 and melted at 900 8C for 1 h in a platinum crucible, followed by quenching and
pulverization to pass 200 mesh sieve. To prepare 0.9MgTiO3–0.1CaTiO3 (MCT) ceramic, the calcined MgTiO3,
CaTiO3, and glass frits were mixed stoichiometrically, ball-milled for 48 h, dried, granulated, and pressed at 98 MPa to
form pellets with 8 mm in diameter and 3 mm in thickness. The pellets were sintered from 900 to 1050 8C for 2 h at a
heating rate of 5 8C/min.
Shrinkage of the specimens during heating was measured using a horizontal-loading dilatometer with alumina
rams and boats (model DIL402C, Netzsch Instruments, Germany). The crystal structure of the sintered sample was
investigated using X-ray powder diffraction (model M18XHF, Macscience Instruments, Japan) in the 2u range from
20 to 608. The polished and thermal etched surfaces of sintered specimens were examined using field emission
scanning electron microscopy (FESEM: model JSM-6330F, JEOL, Japan). The microwave dielectric properties of
the sintered samples were measured using a network analyzer (model HP8720C, Hewlett Packard, USA) in a
frequency range of 8–10 GHz. The quality factor (Q ꢀ f) was measured through the transmission cavity method
using a Cu cavity and Teflon supporter [7]. The relative dielectric constant (k) was measured using the post resonator
method [8] and the temperature coefficient of the resonant frequency (Tf) was measured using a Invar cavity in a
temperature range of 10–80 8C [9].
3. Results and discussion
3.1. Densification
Fig. 1 shows the shrinkage behaviour of the as-pressed specimens with varying LBS glass as a function of rising
temperature. Also, the behaviour of as-pressed LBS glass frit itself is included in Fig. 1 as a reference. The LBS
glass shows a higher rate of shrinkage from about 230 8C, i.e., the glass transition temperature of LBS glass. As the
temperature increases to about 500 8C, another high-shrinkage-rate region (500–540 8C) appears, which results
from the devitrification of the LBS glass. As seen in Fig. 2,1 the LBS glass is crystallized into the Li2B4O7
compound from approximately 500 8C, which may increase the density of the specimen, yielding enhanced
shrinkage from approximately 500 8C as shown in Fig. 1. From about 720 8C, the LBS specimen shows excessive
shrinkage, due to the initiation of the melting of the Li2B4O7 crystals, which melt completely at about 840 8C as
seen in the figure.
According to Fig. 1, a good densification of 0.9MgTiO3–0.1CaTiO3 ceramic requires a firing temperature above
approximately 1300 8C. The increased addition of LBS glass decreases the densification temperature due to the
role of the low-melting compound Li2B4O7. This decrease demonstrates the capability of the new LBS glass
composition as a liquid phase sintering agent for MgTiO3–CaTiO3 based ceramics. As the addition of LBS glass
increases to as high as 20 wt.% (Fig. 1), the shrinkage behaviour of the specimen roughly shows the trend of LBS
itself at temperatures below approximately 650 8C. However, in the 650–7008 C range, additional shrinkage occurs
in the 20 wt.% LBS-added specimen, possibly due to the decomposition of MgTiO3 which will be discussed in the
next section.
The measured apparent densities of sintered specimens with varying amounts of LBS glass are shown in Table 1.2
Although the increase in firing temperature certainly increases the density, in the cases of specimens with relatively
large amounts of glass, e.g., 10 and 20 wt.%, firing at overly high temperatures (e.g., 1050 8C for the 10 wt.%
1
The samples in Fig. 2 were air quenched as the temperature reached the target point.
2
Relative density is hard to determine due to the complicated material reactions described in the next section.

1208
H.-K. Shin et al. / Materials Research Bulletin 41 (2006) 1206–1214
Fig. 1. Shrinkage behaviour of 0.9MgTiO3–0.1CaTiO3 ceramic with varying amounts of LBS glass.
Fig. 2. XRD patterns from lithium borosilicate glass itself heat-treated at varying temperatures. All the peaks are from Li2B4O7.
Table 1
Densities of specimens with varying amounts of LBS sintered at different temperatures (g/cm3)
LBS addition (wt.%)
900 8C
950 8C
1000 8C
1050 8C
5
10
20
2.163
2.225
2.528
2.646
3.383
3.304
3.436
3.552
3.259
3.679
3.487
3.154

H.-K. Shin et al. / Materials Research Bulletin 41 (2006) 1206–1214
1209
Fig. 3. XRD patterns from 0.9MgTiO3–0.1CaTiO3 ceramic sintered with 20 wt.% LBS glass at varying temperatures. The unmarked peaks are from
MgTiO3.
specimen and 1000–1050 8C for the 20 wt.% specimen) slightly lowers the density. This decrease probably occurs
because the volatization of light species in the glass, such as lithium and boron oxides, increase porosity.
3.2. Decomposition of MgTiO3
In order to investigate the material reactions during the elevation of temperatures, the 20 wt.% LBS glass-added
specimens were removed from the furnace during heating (5 8C/min) at varying temperatures for XRD analysis; the
results are shown in Fig. 3. As can be seen in the figure, the decrease in the peak intensity of MgTiO3 (unmarked in
Fig. 3) is apparent with the formation of various secondary phases such as MgTi2O5 (marked as *), Mg2TiO4 (marked
as S), MgO (marked as m), and TiO2 (marked as T). Of the secondary phases, the growth of MgTi2O5 and Mg2TiO4
with the temperature is very significant. Since the decrease in the XRD peak intensity of MgTiO3 (unmarked) and the
increase in Mg2TiO4 (marked as S) take place simultaneously, MgTiO3 is interpreted as decomposing into Mg2TiO4
via the material reaction,
2MgTiO3 ¼ Mg2TiO4 þ TiO2
(1)
In relation to the reaction (1), it is interesting to note that the peak intensity of TiO2 (denoted as T) increases up to
700 8C, while it diminishes at temperatures above 700 8C. However, that of Mg2TiO4 (denoted as S) monotonically
increases, as aforementioned. From this phenomenon, TiO2 is interpreted as dissolving into the LBS glass as the
temperature increases to above approximately 700 8C.
As for the formation of MgTi2O5 (marked as *), it also forms by another decomposition reaction,
2MgTiO3 ¼ MgTi2O5 þ MgO
(2)
Once MgTi2O5 (marked as *) and MgO (marked as m) appear at approximately 750 8C in Fig. 3, the peak intensity of
MgTi2O5 grows continually with temperature. However, that of MgO does not increase correspondingly, especially

1210
H.-K. Shin et al. / Materials Research Bulletin 41 (2006) 1206–1214
Fig. 4. XRD patterns from reference 0.9MgTiO3–0.1CaTiO3 ceramic (sintered at 1300 8C for 2 h) and 0.9MgTiO3–0.1CaTiO3 ceramic sintered with 5,
10, and 20 wt.% LBS glass at 1000 8C for 2 h. MT (MgTiO3) and CT (CaTiO3) symbols are intentionally unmarked in the upper patterns for clarity.
at above 850 8C. From such a limited increase in the peak intensity of MgO, it is also believed to have partially
dissolved into the LBS glass.
In addition to the investigation of the effect of temperature on the material reaction as discussed above, the degree
of material reaction is shown in Fig. 4 as a function of glass addition at 1000 8C. For such purpose, the samples were
held for 2 h at 1000 8C to allow sufficient reaction time. Included in Fig. 4 is the LBS glass-free MCT ceramic,
separately densified at 1300 8C for 2 h as a reference. Similar to the effect of temperature (Fig. 3), the increased
addition of LBS glass is shown to facilitate the decomposition reactions (1) and (2): the peak intensity of MgTiO3
decreases but those of MgTi2O5 and Mg2TiO4 increase with the addition of LBS glass.
However, as apparent in Fig. 4, such decompositions are not observed in the glass-free MCT ceramic composition
fired at 1300 8C. In order to check the thermodynamic feasibility of the decomposition of MgTiO3 itself, Gibbs free
energy change of the reactions (1) and (2) are plotted as a function of reaction temperature in Fig. 5, using
thermodynamic data shown in Barin [10]. In Fig. 5, the Gibbs free energy change of both reactions are positive in the
range of densification temperature: the decompositions of MgTiO3 itself are thermodynamically unfavourable, as
consistent with the XRD peak intensity of the MCT reference shown in Fig. 4.
However, the decomposition reactions actually took place under the presence of the LBS glass. The Gibbs free
energy change (roughly in the range of 16–24 kJ/mol in Fig. 4) for the two reactions is only slightly positive
considering the fact that the Gibbs free energy of formation (DGf) for the compounds involved in the reactions is highly
negative: from 626.85 to 1126.85 8C, DGf for MgTiO3, MgTi2O5, Mg2TiO4, MgO, and TiO2, are ꢁ1308.779 to
ꢁ1158.464, ꢁ2089.79 to ꢁ1855.102, ꢁ1815.309 to ꢁ1610.318, ꢁ504.255 to ꢁ442.919, and ꢁ780.233 to
ꢁ691.634 kJ/mol, respectively [10]. Thus, the appropriate lowering of the activity of the reaction product would
render the Gibbs free energy change for the overall reaction to be negative, in light of the following thermodynamic
relations for reaction Eqs. (1) and (2),
a
a
Mg2TiO4 TiO2
DG ¼ DGꢂ þ RT ln
DG ¼ DGꢂ þ RT ln
¼ DGꢂ þ RT ln aTiO
(3)
(4)
2
a
MgTiO3
a
aMgO
MgTi2O5
¼ DGꢂ þ RT ln aMgO
a
MgTiO3

H.-K. Shin et al. / Materials Research Bulletin 41 (2006) 1206–1214
1211
Fig. 5. Gibbs free energy change for the decomposition reactions as a function of reaction temperature.
where G is Gibbs free energy, a is activity, T is temperature, R is gas constant, and superscript 8 denotes standard state.
In Eqs. (3) and (4), the activity of MgTiO3, MgTi2O5, and Mg2TiO4 was assumed to be in unity as they are pure
condensed phases present in the reaction system. However, the activity (effective mole fraction) of TiO2 and MgO acts
as a variable as these oxides are observed to disappear (dissolve into the LBS glass). Thus, the issue is then how much
the activities of MgO and TiO2 have to be lowered in order to render the Gibbs free energy change of the
decomposition reactions to be negative. The required critical activities to favour the decomposition reactions are
plotted in Fig. 6. As seen in Fig. 6, if the activities of MgO and TiO2 are lowered to 0.049–0.168 and 0.060–0.230,
respectively, within the temperature range 650–1050 8C, the decomposition reactions are thermodynamically
favourable. LBS glass in this work is interpreted to take MgO and TiO2 away from the reaction site (by admitting
them into the glass network), and thereby may lower the activities (effective mole fractions) of MgO and TiO2 from
unity to such values as shown in Fig. 6. Thus, without the role of LBS glass to admit MgO and TiO2 into the glass
network, the decomposition reactions (1) and (2) would be unfavourable.
Fig. 6. Change in critical activity of MgO and TiO2 required to favour the decomposition reactions as a function of reaction temperature.

1212
H.-K. Shin et al. / Materials Research Bulletin 41 (2006) 1206–1214
Fig. 7. SEM micrograph of (a) 0.9MgTiO3–0.1CaTiO3 reference ceramic (sintered at 1300 8C for 2 h) and (b) 20 wt.% LBS glass-added
0.9MgTiO3–0.1CaTiO3 ceramic (sintered at 1000 8C for 2 h).
3.3. Microstructure and dielectric properties
Fig. 7(a) shows the thermally etched microstructure of the MCT reference ceramic fired at 1300 8C for 2 h. The
relatively large grains are MgTiO3 while the relatively smaller ones are either MgTiO3 or CaTiO3, based on energy
dispersive spectroscopy (EDS) analysis (not shown). The thermally etched microstructure of the 20 wt.% LBS glass-
added MCT ceramic fired at 1000 8C for 2 h is shown in Fig. 7(b). The large grains (larger than the undecomposed
MgTiO3 in Fig. 7(a)) marked as ‘a’ are MgTi2O5; and the small grains marked as ‘b’ are Mg2TiO4, based on the EDS
results shown in Fig. 8. Consistent with the XRD results in Fig. 4(a), the MgTi2O5 and Mg2TiO4 grains take the majority
of the volume in the specimen. The small grains having right-angled corners (marked as ‘c’) are shown to be CaTiO3
(Fig. 8). Such grains are often embedded in the larger grains, e.g., MgTi2O5, which might have occurred during the
significantgrowthofthe MgTi2O5 grains. Theneedle-likearea (markedas‘d’) isrich inMg, Ti, andSi, indicatingthatitis
the LBS glass area.3 Although it was difficult to identify individual undecomposed MgTiO3 grains via EDS during the
SEM observation, these grains are believed to take the shape of grains similar to small ones such as Mg2TiO4 (marked as
‘b’). In summary of the microstructural observation, decomposition products MgTi2O5 (large grains) and Mg2TiO4
(smaller ones) are major grains while the decomposition of CaTiO3 is negligible, as consistent with the XRD results.
The sintered specimen, possessing decomposed MgTi2O5 and Mg2TiO4 grains with major crystalline phases,
showed good dielectric properties: the 10 wt.% LBS glass-added sample fired at 950 8C for 2 h showed a permittivity
3
The needle shape may result from the thermal etching during the sample preparation for the SEM.

H.-K. Shin et al. / Materials Research Bulletin 41 (2006) 1206–1214
1213
Fig. 8. EDS results from the areas marked in Fig. 7(b): marked as ‘a’ (a), ‘b’ (b), ‘c’ (c), and ‘d’ (d).
of 19.1, and a quality factor of 13,000 GHz, and a Tf of ꢁ1.3 ppm/8C. In order to investigate the role of the
decomposed products, MgTi2O5 and Mg2TiO4, on the dielectric properties of the overall specimen, pure MgTi2O5 and
Mg2TiO4 phases were prepared separately at 1500 8C for 2 h and 1300 8C for 2 h, respectively (Fig. 9). Although the
major phases are MgTi2O5 and Mg2TiO4 (the permittivities are 17.4 and 14.4, respectively), there exists a very high
permittivity phase CaTiO3 (170) in the densified specimen (Figs. 3, 4, and 7); thus, the permittivity can be higher than
17.4. The quality factors of MgTi2O5 (47,000 GHz) and Mg2TiO4 (55,000 GHz) were superior to that of 0.9MgTiO3–
Fig. 9. XRD patterns from MgTi2O5 and Mg2TiO4 prepared at 1500 8C for 2 h and 1300 8C for 2 h.

1214
H.-K. Shin et al. / Materials Research Bulletin 41 (2006) 1206–1214
0.1CaTiO3 (MCT: 44,000 GHz), explaining the high quality factor (13,000 GHz) of the sintered specimen with the
LBS glass. Hence, the decomposition reaction is not necessarily harmful because of the high dielectric performance of
the decomposition products MgTi2O5 and Mg2TiO4.
4. Conclusions
The decomposition behaviour of MgTiO3 based dielectric ceramic during a liquid phase sintering with lithium
borosilicate glass has been investigated. The presence of the LBS glass results in the decomposition of MgTiO3 via the
reactions, 2MgTiO3 = Mg2TiO4 + TiO2 and 2MgTiO3 = MgTi2O5 + MgO, although they are thermodynamically
unfavourable in glass-free ceramic compositions. Through XRD results and thermodynamic analysis, the role of the
LBS glass is revealed as adopting TiO2 and MgO into the LBS glass network, thereby lowering the activity of TiO2 and
MgO, which, in turn, favours the decomposition reactions. The final microstructure of the MCT dielectric ceramic,
after densification and decomposition reactions, revealed MgTi2O4 and Mg2TiO5 as major grains, which is consistent
with the XRD results. The decomposition reactions were not necessarily harmful because of the high dielectric
performance of the decomposition products, MgTi2O5 (17.4 permittivity; 47,000 GHz quality factor) and Mg2TiO4
(14.4 and 55,000 GHz, respectively).
References
[1] C.Q. Scrantom, J.C. Lawson, IEEE Symposium on Technology for Wireless Application, 1999, pp. 193–200.
[2] K. Wakino, Ferroelectrics 91 (1989) 69–86.
[3] R.C. Kell, A.C. Greenham, G.C.E. Olds, J. Am. Ceram. Soc. 56 (1973) 352–354.
[4] H. Yamamoto, A. Koga, S. Shibagaki, N. Ichinose, J. Ceram. Soc. Jpn. 106 (1998) 339–343.
¨ ¨
[5] H. Jantnunen, R. Rautioaho, A. Uusimaki, S. Leppavuori, J. Eur. Ceram. Soc. 20 (2000) 2331–2336.
¨
¨
[6] H. Jantnunen, A. Uusimaki, R. Rautioaho, S. Leppavuori, J. Am. Ceram. Soc. 85 (2002) 697–699.
[7] D. Kaifez, P. Gullion, Dielectric Resonator, Artech House, Norwood, MA, 1986, p. 327.
[8] B.W. Hakki, P.D. Coleman, IRE Trans. Microwave Theory Tech. 8 (1960) 402–410.
[9] T. Nishikawa, K. Wakino, H. Tamura, H. Tanaka, Y. Ishikawa, IEEE Microwave Theory Tech. Symp. Dig. 3 (1987) 277–281.
[10] I. Barin, Thermochemical Data of Pure Substances, Third ed., VCH, New York, 1995.