Materials Science and Engineering B 164 (2009) 151–155
Contents lists available at ScienceDirect
Materials Science and Engineering B journal homepage: www.elsevier.com/locate/mseb
A low-temperature reactive sintering process for the 5Li2 O–1Nb2 O5 –5TiO2 microwave dielectric ceramics Qun Zeng a,∗ , Wei Li b a
School of Information Photoelectric Science and Engineering, South China Normal University, Guangzhou 510006, China Key Laboratory for Ultrafine Materials of Ministry of Education, School of Materials Science and Engineering, East China University of Science and Technology, Shanghai 200237, China b
a r t i c l e
i n f o
Article history: Received 22 March 2009 Received in revised form 4 August 2009 Accepted 15 August 2009 Keywords: Ceramics Dielectric properties Reactive sintering
a b s t r a c t The B2 O3 -doped 5Li2 O–1Nb2 O5 –5TiO2 composite microwave dielectric ceramics prepared by conventional and low-temperature single-step reactive sintering processes were investigated in the study. Without any calcinations involved, the Nb2 O5 mixture of Li2 CO3 and TiO2 was pressed and sintered directly in the reactive sintering process. More uniform and finer grains could be obtained in the 5Li2 O–1Nb2 O5 –5TiO2 ceramics by reactive sintering process, which could effectively save energy and manufacturing cost. And relatively good microwave dielectric properties of εr = 41, Q × f = 9885 GHz and f = 43.6 ppm/◦ C could be obtained for the 1 wt.% B2 O3 -doped ceramics reactively sintered at 900 ◦ C. © 2009 Elsevier B.V. All rights reserved.
1. Introduction With current tendency towards smaller size, higher reliability, better performance and lower cost, much attention has been paid on the development of low-temperature co-fired ceramics (LTCC) because of their potential application for multilayer integrated circuits (MLICs) in the communication systems [1]. For LTCC materials, in addition to good microwave dielectric properties of high dielectric constant (εr ), high quality factor (Q × f), small temperature coefficient of resonant frequency (f ), they are required to have low cost and good sintering ability below the Ag/Cu melting temperature (961 ◦ C/1064 ◦ C) [2]. Commonly, the LTCC materials are prepared through conventional solid-state reaction process by adding low melting point compounds. Although this process is much cheaper compared with other process such as the wet-chemical process, the cost is still relatively high since two steps of ball-milling and calcinations were always needed and high-energy expense could not be avoided. In order to save the cost, several methods have been found to simplify the processing of ceramics preparation, such as high-energy milling with special equipments and/or reactive sintering [3,4]. As we know, the mixture of the raw materials was sintered directly and the calcinations step or the additional steps of ball-milling could be bypassed in the reactive sintering process. So, the reactive sintering process is considered to be a single-step and economical process. And in recent years, several ceramic materials have been obtained by reac-
∗ Corresponding author. Tel.: +86 20 8163 1628; fax: +86 20 3931 0085. E-mail address:
[email protected] (Q. Zeng). 0921-5107/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.mseb.2009.08.014
tive sintering process, such as PbTiO3 , Pb(Mg1/3 Nb2/3 )O3 ceramics [5,6]. In this article, the simpler single-step low-temperature reactive sintering process was introduced for the 5Li2 O–1Nb2 O5 –5TiO2 (LNT) microwave dielectric ceramics preparation for the first time. And no calcinations and special raw materials/equipments were involved in this reactive sintering process. Our previous study pointed that the 1 wt.% B2 O3 -doped 5Li2 O–1Nb2 O5 –5TiO2 (LNT) ceramics could be densified at around 900 ◦ C by conventional solid-state reaction process and have good microwave dielectric properties of high dielectric constant (∼41), high Q × f values (∼10,000 GHz), low-temperature coefficient [7,8]. However, there are few reports on their details and/or the low-temperature reactive sintering process. The present study details the preparation, microstructures and dielectric properties of the LNT ceramics by low-temperature reactive sintering process. 2. Experimental procedure All samples in this study were prepared from high-purity oxide powders (>99.5%) of Li2 CO3 , Nb2 O5 and TiO2 . Stoichiometric proportions of the above raw materials (5:1:5 by mole) were weighed and then divided into two parts. For the B2 O3 -doped ceramics (denoted as P1 ceramics) preparation by conventional solid-state reaction process, the weighed raw materials were mixed in ethanol using zirconia balls as milling media for 24 h. After drying, the mixtures were calcined at 850 ◦ C for 6 h. Subsequently, 1 wt.% B2 O3 was added into the calcined powder and they were remilled together for 24 h. The powders were then uniaxially pressed under the pressure of 150 MPa into disks measuring 16 mm in diameter and 6–8 mm in
152
Q. Zeng, W. Li / Materials Science and Engineering B 164 (2009) 151–155
thickness. For the B2 O3 -doped ceramics (denoted as P2 ceramics) preparation by low-temperature reactive sintering process, 1 wt.% B2 O3 was directly mixed with the weighed raw materials in ethanol for 24 h. Following the same drying and forming procedures, the P2 ceramic green bodies were obtained. And for the LNT ceramics without B2 O3 , the mixture powder of raw materials was directly pressed into disks. The undoped LNT ceramics, P1 and P2 ceramics were sintered at 840–940 ◦ C for 4 h. All the samples were sintered with the heating rate of 3 ◦ C/min, and cooled with the furnace. The XRD patterns of the sintered ceramics were examined by the X-ray diffraction (XRD) analysis with Rigaku RINT2000 (Cu K␣ radiation generated at 40 kV and 40 mA). The densities of the ceramics were measured by the Archimedes method. The polished and thermally etched surface morphologies were observed with electron probe X-ray microanalyser (EPMA) (JXA-8100). The EPMA samples were polished and thermally etched at about 70 ◦ C below their sintering temperatures for 20 min to reveal their grain structures. The compositional and elemental analysis was carried out with energy dispersive spectroscopy (EDS) (Oxford). The dielectric constant (εr ) and the quality values Q at microwave frequency were measured using the Hakki–Coleman’s dielectric resonator method, as modified and improved by Courtney [9,10]. A vector network analyzer (E8363, Agilent, the U.S.) was employed in the measurement. The temperature coefficient of resonant frequency ( f ) was measured in the temperature range from −25 to +85 ◦ C. The f value was defined as follows: f =
(f85 − f−25 ) × 106 (ppm/◦ C) 110 f25
(1)
where f85 , f−25 , and f25 are the resonant frequencies at 85, −25 and 25 ◦ C, respectively. 3. Results and discussion Fig. 1 shows the X-ray diffraction patterns of the P2 ceramics sintered at 840–900 ◦ C and the P1 ceramic sintered at 900 ◦ C. It could be seen that both the P1 and P2 ceramics were composed of the Li2 TiO3 solid solution (Li2 TiO3 ss) (JCPDS file No. 33-0831) and “M-Phase” phase, no obvious difference could be found in their XRD spectra [11,12]. In addition, estimating from the XRD results, the Li2 TiO3 ss/M-phase volume ratios of the P2 ceramics sintered at different temperatures keep almost unchanged, which implies that the phase formation has almost completed at temperature lower
Fig. 1. The X-ray diffraction patterns of the P2 ceramics sintered at 840–900 ◦ C and the P1 ceramic sintered at 900 ◦ C.
Fig. 2. The bulk densities of P1, P2 and the LNT ceramics without B2 O3 vs. sintering temperature.
than 840 ◦ C. In addition, in the case of P2 samples, it could be seen that XRD peak shift to higher angle side with increasing sintering temperature. The probable reason for the peak shift in Fig. 1 might be the uncertain reaction between “M-Phase” and Li2 TiO3 ss phase as the temperature increasing [13] and it induced the distortion of the crystal lattice. And the exact reason is not so clear yet and still under investigation. In reference [14,15], there were also XRD peak shift phenomenon in Li2 O–Nb2 O5 –TiO2 system ceramics, and the reason was not referred. Fig. 2 shows bulk densities of P1, P2 and the LNT ceramics without B2 O3 as a function of sintering temperature from 840 to 940 ◦ C. And the powder which is used for sintering of the LNT ceramics without B2 O3 is not calcined. As shown in Fig. 2, the P1 samples achieve a relatively high density when the sintering temperature reached 860 ◦ C. And with the sintering temperature increasing to 920 ◦ C, there is a slight decrease in the density. On the other hand, it could be observed that the densities of the P2 ceramics increased with the increasing temperature and the relatively high densities could be obtained at about 920 and 940 ◦ C. In addition, it could be seen that the densities of the undoped LNT ceramics were much lower than those of P1 and P2 ceramics. It is obvious that the LNT ceramics without B2 O3 could not be sintered well at lower than 940 ◦ C. Moreover, even though the powders used for sintering of the LNT ceramics without B2 O3 are calcined in advance, the LNT ceramics without B2 O3 also could not be sintered densely at lower than 1000 ◦ C (the density of the pure ceramics sintered at 950 ◦ C is about 3.38 g/cm3 ). In Reference 7, it was reported that the sintering temperature of dense undoped LNT ceramics should be about 1100 ◦ C. That is to say, the density of the pure LNT ceramics does not depend on the power calcination. So, it could be concluded that B2 O3 is also a good sintering aid for the reactive sintering process and very effective in enhancing the sintering ability of the LNT ceramics. Moreover, as discussed in Fig. 1, the phase formation of the P2 ceramics has already completed at temperature lower than 840 ◦ C, so the density difference between P1 and P2 ceramics are not due to phase transformation or the chemical reaction during sintering. And the volume changes, internal stress generation during phase transformation and the expelled gas CO2 are assumed to be the reasons for the lower densities of P2 ceramics. In order to make clear which factor is primary, the ceramic pellets are also prepared from prefired (750 ◦ C) powder of Li2 CO3 , i.e., the raw material Li2 CO3 is substituted by powder of Li2 O. The densities of the pellets by conventionally sintering method are similar with those of P1 ceramics, whereas the densities of these pellets by reactive sintering process are higher than those of P2 ceramics at 840–900 ◦ C.
Q. Zeng, W. Li / Materials Science and Engineering B 164 (2009) 151–155
153
Fig. 3. SEM micrographs of (a–c) P1 ceramics sintered at 860–900 ◦ C and (d–g) P2 ceramics sintered at 860–920 ◦ C.
So, the expelled gas CO2 is considered to be the primary factor for the lower densities of P2 ceramics. The adverse effect of expelled CO2 on the density is also referred in reference [16]. The SEM micrographs of (a–c) the P1 ceramics sintered at 860–900 ◦ C; (d–g) P2 ceramics sintered at 860–920 ◦ C are illustrated in Fig. 3. From the SEM micrographs, it could be observed that sample P1 and P2 have similar microstructures, i.e., both of them are composed of two kinds of grains, one is bright (denoted as A) and another one is dark (denoted as B). Combined EDS with XRD results, the dark grains B could be considered as the Li2 TiO3 ss, while bright grains A are considered as M-phase. Besides, the grain sizes
of both P1 and P2 ceramics increase as the sintering temperature rising. Nevertheless, the differences between the microstructures of P1 and P2 ceramics are also obvious. Compared with the P1 ceramics, at the temperature of lower than 900 ◦ C, there are more pores in the P2 ceramics. In addition, at the same sintering temperatures, it could be obviously found that the grain sizes of the P2 ceramics were more uniform and finer than those of P1 ceramics. The quantitative analysis results of the average grain size and standard deviations data have been shown in Fig. 4. The grain sizes of Li2 TiO3 ss in P1 and P2 ceramics were determined on polished, etched surfaces using a linear intercept technique [17]. From
154
Q. Zeng, W. Li / Materials Science and Engineering B 164 (2009) 151–155
Fig. 5. The microwave dielectric constants of P1 and P2 ceramics as a function of sintering temperature.
Fig. 4. Grain sizes of Li2 TiO3 ss in P1 and P2 ceramics as a function of sintering temperature.
Fig. 4, it could be observed that the average grain sizes of Li2 TiO3 ss increase with sintering temperature. And the standard deviation data for P1 ceramics are a bit larger than those for P2 ceramics. Moreover, in Fig. 3(b) and (c), it could be seen that some abnormal abnormally grown large grains of M-phase existed. Combined Figs. 3 and 4, it might be concluded that the grain size distribution of the P2 ceramics is more uniform and finer than that of P1 ceramics. The phase compositions of the sample P1 and P2 sintered at 900 ◦ C qualitatively identified by EDS (Energy Dispersive Spectroscopy), as presented in Table 1. No obvious variation of Nb/Ti ratios of the M-phase and Li2 TiO3 ss in P1 and P2 ceramics could be observed from Table 1. It suggests that the phase compositions of sample P1 and P2 are very similar though they were prepared by different process.
The microwave dielectric constants of P1 and P2 ceramics as a function of sintering temperature are shown in Fig. 5. Generally, the dielectric constant of the sintered ceramics is determined, to an extent, by the bulk densities of the sintered ceramics, due to the low εr value of pores (∼1.0). Comparing Figs. 2 and 5, the similar trend could be found, i.e., the curves of the εr value vs. temperature are similar as those of density vs. temperature. For the P1 ceramics, the εr values first increase slightly with the sintering temperatures and achieve a relatively high value when the sintering temperature reached 880 ◦ C, then the εr values decrease slightly with further increasing sintering temperature up to 920 ◦ C due to their lower densities. And for the P2 ceramics, the εr values increase with the sintering temperatures and higher dielectric constant was obtained at 920 ◦ C. The highest dielectric constant (εr ) of ∼41 could be obtained at 920 ◦ C. Fig. 6 shows the Q × f values of P1 and P2 ceramics as a function of sintering temperature. It could be found that the Q × f values of P2 ceramics keep increasing as the sintering temperature rising, and the Q × f values of the P2 ceramics are a bit higher than those of P1 ceramics at the same sintering temperatures, although the densities of the P2 samples are lower than P1 samples. As we know, the microwave dielectric loss includes not only intrinsic losses which were mainly caused by the lattice vibrational modes but also extrinsic losses dominated by the secondary phases, densifica-
Table 1 The phase compositions of the sample P1 and P2 sintered at 900 ◦ C. Sample
A (M-Phase) Atomic%
P1 P2
B (Li2 TiO3 ss phase) Ratios
Atomic%
Ratios
Nb
Ti
Nb/Ti
Nb
Ti
Nb/Ti
12.01 12.11
19.36 19.25
0.620 0.629
6.18 6.06
25.30 25.9
0.244 0.233
Fig. 6. The Q × f values of P1 and P2 ceramics as a function of sintering temperature.
Q. Zeng, W. Li / Materials Science and Engineering B 164 (2009) 151–155
155
Due to the similar phase compositions in P1 and P2 ceramics, the f values of P1 and P2 ceramics did not have much difference in this experiment. 4. Conclusions
Fig. 7. Variation of the resonant frequency of P1 and P2 ceramics vs. the sintering temperature.
tion/porosity, oxygen vacancies, grain sizes, and grain morphology and uniformity [18]. Some investigations [19,20] also reported that the Q × f value is independent of the density and the porosity for a theoretical density higher than 90%. As for the P1 and P2 ceramics, their densities are higher than 90% and no other phases could be found, so the grain sizes, grain morphology and uniformity are suggested to dominate the Q × f values. Generally, on the condition of the same grain uniformity, as the average grain size increased, the number of grain boundaries per unit volume decreased, i.e., the sources of loss decreased; therefore, the dielectric loss would generally decrease [21]. From Figs. 3 and 4, it could be found that the grain size of P1 and P2 ceramics increased with the increasing sintering temperature, therefore, the Q × f values might increase with the increasing temperature. However, another interesting phenomenon could be also observed. The Q × f values of P2 ceramics are higher than those of P1 ceramics, although the grains of P2 are much finer. This phenomenon might be caused by the different grain morphology and uniformity. In Fig. 3 it could be observed that the grains of P2 are more uniform than those of P1. In addition, from Fig. 4, it could be found that the uniformity of grains in P2 ceramics is better than that of P1 ceramics, although the average grain size of P2 is lower. Also, it could be seen that there are several abnormally grown large “M-phase” grains in P1 ceramics as shown in Fig. 3(b) and (c). So, it is considered that the grain morphology and uniformity might be the primary factor for the Q × f values. Therefore, it might be not difficult to understand that the Q × f values of P2 are higher than P1. The relatively maximum Q × f value of 10,119 GHz could be obtained for the P2 ceramics at 920 ◦ C. The variations of resonant frequency in the TE011 mode of the P1 and P2 ceramics as a function of the temperature are shown in Fig. 7. The calculated f values of the P1 and P2 ceramic sintered at 900 ◦ C are 45.5 and 43.6 ppm/◦ C, respectively. It is well known that the temperature coefficient of resonant frequency ( f ) is related to the composition and the secondary phase of the materials [22].
The B2 O3 -doped 5Li2 O–1Nb2 O5 –5TiO2 (LNT) composite microwave dielectric ceramics has been prepared by conventional and low-temperature single-step reactive sintering processes. For the reactive sintering, the raw materials are sintered directly and no calcinations are involved. So the manufacturing cost and energy could be effectively saved. And even the calcinations stage is bypassed, the LNT ceramics (P2 ceramics) obtained by reactive sintering process could also have good microwave dielectric properties as P1 ceramics by conventional solid-state reaction process. Especially the Q × f values of P2 ceramics are a bit higher than those of P1 ceramics at the same sintering temperatures, which is probably due to the more uniform microstructure in P2 ceramics. For the 1 wt.% B2 O3 -doped LNT ceramics reactively sintered at 900 ◦ C, good microwave dielectric properties of εr = 41, Q × f = 9885 GHz and f = 43.6 ppm/◦ C could be obtained. Acknowledgments The authors acknowledge Professor Guo Jingkun and Shi Jianlin of Shanghai Institute of Ceramics Chinese Academy of Science for their support of the testing experiments and helpful discussions. References [1] H. Jantunen, R. Rautioaho, A. Uusimäki, S. Leppävuori, J. Eur. Ceram. Soc. 20 (2000) 2331–2336. [2] T. Takada, S.F. Wang, S. Yoshikawa, et al., J. Am. Ceram. Soc. 77 (1994) 2485–2488. [3] Y.C. Liou, C.Y. Liu, Mater. Sci. Eng. A 448 (2007) 351–355. [4] Y.C. Liou, W.H. Shiu, C.Y. Shih, Mater. Sci. Eng. B 131 (2006) 142–146. [5] J.H. Chen, Y.C. Liou, K.H. Tseng, Jpn. J. Appl. Phys. 42 (1A) (2003) 175–181. [6] Y.C. Liou, C.Y. Shih, C.H. Yu, Mater. Lett. 57 (2003) 1977–1981. [7] Q. Zeng, W. Li, J.L. Shi, J.K. Guo, Mater. Lett. 60 (2006) 3203–3206. [8] Q. Zeng, W. Li, J.L. Shi, X.L. Dong, J.K. Guo, Phys. Stat. Sol. (a) 204 (10) (2007) 3533–3537. [9] B.W. Hakki, P.D. Coleman, IRE Trans. MTT 8 (1960) 402–410. [10] W.E. Courtney, IEEE Trans. MTT 18 (1970) 476–485. [11] Powder Diffraction File (PDF) No. 33-0831. [12] A.Y. Borisevich, P.K. Davies, J. Am. Ceram. Soc. 85 (3) (2002) 573–578. [13] Q. Zeng, W. Li, J.L. Shi, X.L. Dong, J.K. Guo, J. Am. Ceram. Soc. 91 (2) (2008) 644–647. [14] D.H. Kang, K.C. Nam, H.J. Cha, J. Eur. Ceram. Soc. 26 (2006) 2117–2121. [15] H.F. Zhou, H. Wang, D. Zhou, L.X. Pang, X. Yao, Mater. Chem. Phys. 109 (2008) 510–514. [16] A.Y. Borisevich, D.K. Davies, J. Am. Ceram. Soc. 87 (6) (2004) 1047–1052. [17] L.C. Stearns, M.P. Harmer, J. Am. Ceram. Soc. 79 (12) (1996) 3013–3019. [18] B.D. Silverman, Phys. Rev. 125 (1962) 1921–1930. [19] W.S. Kim, T.H. Hong, E.S. Kim, K.H. Yoon, Jpn. J. Appl. Phys. 37 (1998) 5367–5371. [20] C.L. Huang, M.H. Weng, Mater. Res. Bull. 35 (2000) 1881–1888. [21] S.H. Yoon, G.K. Choi, D.W. Kim, S.Y. Cho, K.S. Hong, J. Eur. Ceram. Soc. 27 (2007) 3087–3091. [22] C.L. Huang, K.H. Chiang, S.C. Chuang, Mater. Res. Bull. 39 (2004) 1701–1708.