Effect of additives on the sintering of MgAl2O4

Effect of additives on the sintering of MgAl2O4

Journal of Alloys and Compounds 587 (2014) 594–599 Contents lists available at ScienceDirect Journal of Alloys and Compounds journal homepage: www.e...

1MB Sizes 3 Downloads 95 Views

Journal of Alloys and Compounds 587 (2014) 594–599

Contents lists available at ScienceDirect

Journal of Alloys and Compounds journal homepage: www.elsevier.com/locate/jalcom

Effect of additives on the sintering of MgAl2O4 Taehyung Kim, Donghyun Kim, Shinhoo Kang ⇑ Department of Materials Science and Engineering, Seoul National University, Seoul 151-742, Republic of Korea

a r t i c l e

i n f o

Article history: Received 14 July 2013 Received in revised form 22 October 2013 Accepted 31 October 2013 Available online 8 November 2013 Keywords: Spinel Sintering Additive Hardness Structural ceramics

a b s t r a c t The effect of additives on the densification of stoichiometric spinel was investigated. SiO2, CaCO3 and TiO2 are used as additives to spinel composition in the range of 1–4 wt%. Densification behavior is investigated by dilatometry measurement and the sintered body is characterized in terms of microstructure, phase formation and change in lattice parameter. In case of SiO2 and CaCO3 additions, the densification was significantly enhanced with the additives by forming glassy phases in grain boundary region. The addition of TiO2 resulted in the formation of secondary phase at grain boundaries and inside grains along with enhanced densification. Ó 2013 Elsevier B.V. All rights reserved.

1. Introduction Magnesium aluminum oxide, the mineral name of which is spinel, is a well-known refractory material with its excellent properties: high resistance to chemical attack, high temperature mechanical property and good thermal shock resistance [1,2]. For the synthesis of the spinel, conventional solid state reaction is used with Al2O3 and MgO as raw materials. The formation mechanism is known as the interdiffusion of Al3+ and Mg2+ cations through the oxide particles and the reaction temperature is relatively high (>1400 °C) [3]. In addition, due to the large volume expansion of more than 5% during the spinel formation, it is difficult to make dense spinel sintered body by a single step sintering [4]. As for the densification of the single phase Mg spinel, it is known that densification behaviors are different as a function of the degree of nonstoichiometry. MgOrich compositions are known to be sintered more rapidly and achieved higher density compared with Al2O3 excess ones [5]. The rate-controlling mechanism for the sintering of nonstoichiometric MgAl2O4 compositions is suggested as oxygen lattice diffusion through oxygen vacancies [5]. Effects of additives on the sintering of stoichiometric or nonstoichiometric Mg spinel are also reported for various additives. In alumina-rich spinel, the dissolution and precipitation of alumina was observed with the addition of LiF and CaCO3, forming low temperature compounds [6–8]. SiO2 is known to form various silicatebased glass materials with Al2O3 or Al2O3-based ceramics in the grain boundaries. Also, rare earth oxides (Yb2O3 and Dy2O3) were ⇑ Corresponding author. Tel.: +82 2 880 7167; fax: +82 2 884 0074. E-mail address: [email protected] (S. Kang). 0925-8388/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jallcom.2013.10.250

also reported to improve densification of spinel. The effect of the rare earths on sintering is explained by the crystallochemical characteristics of the cations which increase the activity of rare-earth oxides in forming cation vacancies [9]. Y2O3 was reported to enhance the densification of stoichiometric and MgO-rich spinels [10]. The addition of TiO2 was found to enhance densification by exsolution of alumina and dissolution of TiO2 [10]. In case of ZnO addition, the formation of anion vacancy was suggested to help densification of spinel [11]. These various results about the densification of spinel are closely related to the unique property of Mg spinel phase. According to the MgO–Al2O3 binary phase diagram in Fig. 1, there are relatively high solubility of MgO and Al2O3 in the spinel phase. It has much more solubility of MgO compared to that of pure Al2O3, which is another important fact in sintering study [5]. The mutual high solubility means that the spinel phase can accommodate related defects like cation or oxygen vacancies. Therefore, for the different kinds of additives, the reaction involving these defects is one of the major factors determining the densification and coarsening process. In this study, we have investigated the densification behavior of stoichiometric spinel with the addition of 1–4 wt% CaCO3, SiO2 and TiO2 for the structural applications of Mg spinel at ambient temperature. The physical and microstructural changes and mechanical properties are also examined and discussed in terms of phase formation and compositional variations. 2. Experimental The raw materials selected for this study were high purity a-Al2O3 (99.9%, Kojundo, Japan) and MgO (99.9%, Kojundo, Japan) powders. To make stoichiometric spinel power, conventional powder mixing and solid state reaction was used. The

595

T. Kim et al. / Journal of Alloys and Compounds 587 (2014) 594–599 2.0 1.8

1400

1.6

1200

1.4 1.2

1000

1.0

800

0.8

600

0.6

400

0.4

200

0.2

dl/ (dtLo)

temperature (oC)

1600

0.0

0

-0.2 100

120

140

160

180

200

220

time (min) Fig. 2b. Densification rate of SiO2-added spinel. Fig. 1. Phase diagram of MgO–Al2O3 system.

3. Results and discussion

dl/Lo (%)

dl/Lo (%)

temperature (oC)

Fig. 2a shows the shrinkage curves of spinel with SiO2 (1–4 wt%) as a function of time. The sintering schedule is shown together. It can be seen that the addition of SiO2 greatly promotes the densification of spinel. The final value of the relative shrinkage (%DL/L0) is more than three times that of pure spinel. The additive amount also changed the densification behavior. Fig. 2b shows the densification rate curves, which is the derivative of the curves in Fig. 2a. In the figure it is apparent that SiO2-added samples have much higher densification rates than pure spinel. Further, the densification starts at a lower temperature and finishes faster as the added amount of SiO2 increases.

The addition of CaCO3 had a strong effect on the densification of spinel. The densification curves of CaCO3-added samples are shown in Fig. 3a. The relative shrinkages of CaCO3-added samples are also much improved and the results are similar to those of the SiO2added spinel except the fact that the system with CaCO3 has a maximum shrinkage rate at a lower temperature (1350 °C) than that of SiO2-added system (1400 °C). Fig. 3b shows that densification rate increases and the time for densification decreases as the amount of CaCO3 increases. However, the change in the initiation of notable sintering is not continuous as a function of additive content in this case. The densification curves of TiO2-added samples and pure spinel are shown in Fig. 4. It is interesting to find that the maximum values of relative shrinkage are similar for all the additives, which are summarized in Table 1. The position and height of the peaks from TiO2-added samples are almost the same for the samples with different amounts of TiO2. However, the peak widths are wider than those with SiO2 and CaCO3. The observation might be related to the forming tendency of the grain boundary phases and the reaction kinetics. It is known that TiO2 tends to crystallize the grain boundary phase while others would form a glassy phase or a low-temperature compound at the boundaries [8]. This is proven in their microstructures that will be discussed later. In addition, it is noted that the densification of TiO2-added samples started at a lower temperature: 1100 °C for the samples with TiO2 and 1200 °C for the samples with SiO2 and CaCO3. It might be attributed to the difference in the formation temperature of the phases initially forming in the systems.

temperature (oC)

oxide powders were mixed in Al2O3:MgO = 1:1 ratio by ball milling in ethanol media. After drying the mixed powders were calcined at 1350 °C for 2 h, which results in the formation of pure spinel. 1–4 wt% additives such as SiO2 (99.9%, Kojundo, Japan), CaCO3 (99.9%, Kojundo, Japan) and TiO2 (99.9%, Aldrich, USA) are mixed with the synthesized spinel powder by ball milling. Dried powder was uniaxially pressed at 125 MPa to make 10 mm diameter disk and sintered at 1650 °C for 2 h in air. For the characterization of the densification behavior, linear shrinkage was measured by a dilatometer (DIL402, NETZSTH, Germany) and bulk density of the sintered body was measured by the Archimedes principle. The sintered specimens were mechanically ground and polished with alumina powder. Phase analysis of the samples was carried out by XRD analysis with a Rigaku D-Max2500 diffractometer equipped with Cu Ka (k = 1.54056 Å) radiation source and a rotating anode. Microstructural analysis was done in a SEM (JMS-6360, Japan Electronic Optics Laboratory, Japan) attached with electron diffraction X-ray analysis (EDXA) facility. For the observation of the grain structure of pure spinel and spinel with additives, thermal etching was done at 1500 °C for 4 and 2 h, respectively.

time (min) time (min) Fig. 2a. Dilatometric shrinkage curve of pure spinel and SiO2-added spinel. Solid and dotted lines are for temperature and shrinkage, respectively.

Fig. 3a. Dilatometric shrinkage curve of pure spinel and CaCO3-added spinel.

T. Kim et al. / Journal of Alloys and Compounds 587 (2014) 594–599

1600

1200 1000

3.50

1.8

3.48

1.6

3.46

1.4

3.44

1.2 1.0

800

0.8

600

dl/ (dtLo)

temperature (oC)

1400

2.0

0.6

Density (g/cm3)

596

3.42 3.40 3.38 3.36

400

0.4

3.34

200

0.2

3.32

0.0

3.30

0

-0.2

80

100

120

140

160

180

200

220

TiO

2

CaCO

3

SiO

2

3.28 0

240

1

time (min)

temperature (oC)

dl/Lo (%)

time (min) Fig. 4a. Dilatometric shrinkage graph of pure spinel and TiO2-added spinel.

1.0 1600 0.8

1200 0.6

1000 800

0.4

600 400

dl/ (dtLo)

temperature (oC)

1400

0.2

200 0.0

0 100

120

140

160

180

200

220

240 3.8

Fig. 4b. Densification rate of TiO2-added spinel.

3.6

TiO

3.4

SiO

2 2

CaCO

3

Table 1 Maximum relative shrinkages ( 100*dl/L0) of pure spinel and others with additives.

pure spinel SiO2 CaCO3 TiO2

6.19

1 20.37 21.36 23.10

2 22.46 21.76 22.24

3 23.08 22.86 22.38

4 23.63 22.86 22.04

Density (g/cm3)

3.2

0

4

sintering temperature and time were set at 1650 °C and 2h, respectively, for these samples. This sintering condition is found to provide full densification of pure spinel as shown in Fig. 6 and other samples with additives also show the maximum bulk densities from this condition. Upon the addition of 1 wt% additive, there is an increase in bulk density for all three samples compared to pure spinel in Fig. 5. However, as the amount of additive increases, the SiO2- and CaCO3-added samples show a decrease in density in contrast to the shrinkage results shown in Figs. 2 and 3. This is attributed to the low density values of SiO2 and CaO, i.e., 2.65 and 3.3 g/cm3, respectively compared to that of pure spinel, 3.6 g/cm3. Thus, further additions of these additives result in low density values even with increase in shrinkages. But TiO2-added one shows maximum around 3 wt% addition due to its effectiveness in densification and to its density value, 4.2 g/cm3. The optimum amounts of additives, when determined by the density values, are found at 1 wt% additions for SiO2 and CaCO3 and 3 wt% for TiO2. The effects of the 3 wt% additives on the bulk density are shown in Fig. 6 with respect to sintering temperature. The samples with the additives provide much higher density at a lower temperature than pure spinel. The density of pure spinel increases slowly as sintering temperature but there is a sudden increase around 1300– 1400 °C for the samples with additives. In the case of TiO2 addition, full density is obtained at the lowest temperature (1400 °C) among the all samples. In the view of the densification rate and density values, TiO2 is found as the most effective additive. Fig. 7 is SEM images of fractured surfaces of various spinel with and without additives, 3 wt%, after sintering at 1650 °C for 2h. Pure

time (min)

Additives (in wt%)

3

Fig. 5. Bulk density changes as a function of the amount of additives (sintered at 1650 °C for 2 h).

Fig. 3b. Densification rate of CaCO3-added spinel.

80

2

Additive content (wt%)

spinel

3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6 1200

The bulk densities of those samples are measured as a function of the amount of various additives and shown in Fig. 5. The

1300

1400

1500

1600

Temparature (oC) Fig. 6. Bulk density changes as a function of additives and sintering temperature.

T. Kim et al. / Journal of Alloys and Compounds 587 (2014) 594–599

597

Fig. 7. SEM images of fracture surface of (a) pure spinel, (b) CaCO3-, (c) SiO2- and (d) TiO2-added spinels.

spinel shows grains of 5–10 lm in size with many pores without much grain boundary glassy phase. From fractured surfaces, the mixed mode of intra- and inter-granular fracture is observed as shown in Fig. 7(a). But the grain boundaries are more clearly seen in the fracture surfaces with additives. It means that the additives, in general, enhance the inter-granular fracture of spinel by forming a brittle phase which could be path to crack propagation. Pure spinel demonstrates a fine grain structure while 3 wt% CaCO3-added sample in Fig. 7(b) shows a larger grain size and number of pores than pure spinel sample. The pore size is in the range of 1–10 lm. The addition of 3 wt% SiO2 results in much larger grains of 5–20 lm in average and some pores of >10 lm located at the grain boundary junctions. Fig. 7(d) shows the largest grain size of >100 lm with TiO2 and small pores of <5 lm exist at the grain boundaries, indicating that a significant grain growth occurred during sintering stage. In comparison with the SiO2- and CaCO3added samples, the rate for the grain growth in the TiO2-added sample is much fast that the grains tend to trap the pores coalesced at the grain boundaries. The characteristics of the grain boundary phases of each system are more clearly represented by the polished surfaces that were exposed to thermal etching. Fig. 8(a) shows that the grain size is 5–10lm and most pores exist at the grain boundaries for the pure spinel sample. In the SiO2-added sample, it can be seen that the grain size is increased to be 10–20lm and the number of pores decreased as shown in Fig. 8(b). These are similar results observed in the fractured surface images. The thermal etching enlarged the grain boundary regions significantly with SiO2-added sample. This result suggests that SiO2 exists mainly in the form of a glass phase in the grain boundaries, which is proven with a separate XRD analysis. The CaCO3-added sample of Fig. 8(c) shows a similar microstructure to that of SiO2-added sample. With CaCO3 as an additive, there is no evidence of CaAl2O4 and any other Ca-compounds of low melting points as reported [7,8] But our XRD result, not shown in this paper, indicates clearly the presence of CaAl4O7 in the 3 wt% CaCO3-added system, which is known to form above 1250 °C. The formation of CaAl4O7 might be due to the high sintering

temperature of this study, 1650 °C, and the phase seems to have a relatively high stability in contrary to the free energy of formation reported. [8] In a separate study with element mapping, Si and Ca elements are found to be distributed in the grain boundary regions. The addition of TiO2 results in a quite different microstructure. The grain size is much bigger than the other samples as observed in Fig. 7, but many secondary particles are clearly seen both in the grain boundary and grain interior. The etched area is quite narrow to show only grain boundaries as in the case of pure spinel. It shows that the TiO2 addition induces to form the secondary phase with spinel. This phase was identified as Al2TiO5 by XRD analysis. The formation reaction of this oxide phase is likely to enhance the densification of TiO2-added spinel sample at a relatively low temperature. It is yet to be investigated how this phase in a rod shape enhance the densification of spinel structure. It shows that the densification rate is not a strong function of the TiO2 content as shown in Fig. 4. The effect of the additives on the lattice parameter of spinel was investigated from the X-ray diffraction patterns. Fig. 9 shows the measured lattice parameters of pure spinel sample and of those with additives sintered at 1650 °C for 2h. The lattice parameter of pure spinel was 8.078 Å, which is similar to the reported value in JCPDS database. In contrast, the lattice parameters of all samples with additives are small compared to that of pure spinel. Furthermore, the degree of reduction in the lattice parameter varies with respect to the kind of additives. In the case of SiO2 and CaCO3 addition, the lattice parameters remain almost constant regardless of the additive contents. In the TiO2 addition, however, the lattice parameter decreases in proportion to the amount of TiO2. All cations from the additives have larger ionic radii than Mg2+ and Al3+ ions in spinel. Thus, it seems contradictory that all additives reduce the parameters if complete substitution of the additive elements is assumed. It is also known that lattice parameter of spinel decreases both in MgO- and Al2O3-rich non-stoichiometric composition [12]. Therefore, some possible causes that explain the defect formation in relation to the reduction of lattice parameters might be: (1) the incorporation of additive elements

598

T. Kim et al. / Journal of Alloys and Compounds 587 (2014) 594–599

Fig. 8. SEM images (BSE mode) of the surface of the etched samples – (a) pure spinel, (b) SiO2-spinel, (c) CaCO3-, and (d) TiO2-added spinels.

lattice parameter (A)

8.08 pure spinel TiO2 added spinel CaCO3 added spinel SiO2 added spinel

8.06

8.04

The addition of SiO2 or CaCO3 brings about a large volume of a glassy phase or low-melting compound in the grain boundary regions (Fig. 8), indicating their preferential reactions with spinel. The reactions might have resulted in nonstoichiometric spinel. However, previous study reported that there is no effect of CaCO3 addition on the spinel stoichiometry for the system with CaAl2O4 [7]. Thus, it is likely with SiO2 or CaCO3 as a sintering additive that the formation of a low-melting grain boundary phase has little influence on the overall spinel stoichiometry along with limited solubility of SiO2 and CaO in the spinel composition, causing little change in their lattice parameters.

8.02

4. Conclusions 1

2

3

4

wt% Fig. 9. Changes in the lattice parameters as a function of the amount of additives.

in the spinel matrix and (2) the change in the stoichiometry via the phase formation such as secondary and grain boundary phases. Then, the degree of the reduction would be determined by the solubility of an additive to spinel and/or by the reactivity of the additive with spinel. The Si4+ and Ti4+ ions are known to be incorporated in the spinel lattice by producing cation vacancies [13–15]. Oxygen vacancies are primarily associated with entropy of mixing at high temperature for equilibrium. Hence, Schottkey pairs are expected to form if Si4+ or Ti4+ ions take Mg2+ tetragonal sites, resulting in Mg2+ vacancies. Recent study on Ti-doped Mg spinel demonstrated that Ti+3 and Ti+4 ions can take tetrahedral and octahedral sites, substituting Mg+2 and Al+3 in the spinel structure and resulting in increase of Schottkey defects [16]. If the additives react extensively with spinel matrix, it would change the stoichiometry appreciably. The strong tendency to form TiAl2O5 is also expected to influence the stoichiometry of the spinel. Thus, the decreasing behavior of lattice parameter in TiO2-added sample should be explained not only by high solubility of TiO2 in the spinel lattice, but also by the formation of the secondary phase, TiAl2O5, in Fig. 8(d).

The additions of SiO2, CaCO3 and TiO2 to the stoichiometric spinel are found to improve the densification of spinel. Of these additives TiO2 is the most effective in densification. Based on the microstructure and lattice parameter analysis, the densification is found to be different for each additive. In case of SiO2 and CaCO3 additions, the most additive components exist in the form of grain boundary phase at the grain boundaries whereas TiO2 formed the TiAl2O5 phase not only in the grain boundary region and but also in the grain interior through the reaction with spinel. The characteristics of second phase formed and the solubility of each additive in the spinel are presumed to affect the lattice parameters of various spinel. Acknowledgements This work was supported partially by ROTEM Ceramic program (2005-2008). We acknowledge the use of facilities at the Research Institute of Advanced Materials, Seoul National University. We also acknowledge the assistance from J.H. Kim and S.W. Nam for the manuscript preparation and experiment, respectively. References [1] J.H. Chesters, Refractories Production and Properties, The iron and steel institute, London, 1973. [2] K. Shaw, Refractories and their Uses, Applied Science Publishers, London, 1972.

T. Kim et al. / Journal of Alloys and Compounds 587 (2014) 594–599 [3] [4] [5] [6] [7] [8]

R.E. Carter, J. Am. Ceram. Soc. 44 (3) (1961) 116–120. E. Carter, J. Am. Ceram. Soc. 44 (3) (1965) 116–120. C.J. Ting, H.Y. Lu, J. Am. Ceram. Soc. 82 (4) (1999) 841–848. J.L. Huang, S.Y. Sun, Mater. Sci. Eng. A 259 (1999) 1–7. J. Aguilar, A. Arato, M. Hinojosa, U. Ortiz, Mat. Sci. Forum 442 (2003) 79–84. A.H. De Aza, P. Pena, M.A. Rodriguez, R. Torrecillas, S. De Aza, J. Eur. Ceram. Soc. 23 (2003) 737–744. [9] L.A. Skomorovskaya, Glass Ceram. 50 (4) (1993) 165–168. [10] R. Sarkar, S.K. Das, Ceram. Int. 29 (2003) 55–59. [11] A. Ghosh, S.K. Das, Ceram. Int. 26 (2000) 605–608.

599

[12] Y.M. Ching, W.D. Kingery, Phys. Ceram. Handbook (1997) 105–115. [13] A. Jouini, A. Yoshikawa, A. Brenier, T. Fukuda, G. Boulon, Phys. Status Solidi C 4 (2007) 1380–1383. [14] Y. Fujimoto, H. Tanno, K. Izumi, S. Yoshida, S. Miyazaki, M. Shirai, K. Tanaka, Y. Kawabe, E. Hanamura, J. Lumin. 128 (2008) 282–286. [15] T. Sato, M. Shirai, K. Tanaka, Y. Kawabe, E. Hanamura, J. Lumin. 114 (2005) 155–161. [16] J.H. Lim, B.N. Kim, Y. Kim, S. Kang, R.J. Xie, I.S. Chong, K. Morita, H. Yoshida, K. Hiraga, Appl. Phys. Lett. 102 (2013) 031104.