Effects of Additives on the Perovskite Formation in Sol-Gel Derived Lead Magnesium Niobate

(1)

Effects of additives on perovskite formation in sol±gel

derived lead magnesium niobate

W.-F.A. Su

*

Institute of Materials Science and Engineering, National Taiwan University, Taipei, Taiwan Received 9 March 1999; received in revised form 24 May 1999; accepted 28 May 1999

Abstract

The effects of ionic compounds (PbO, BaTiO3, PbTiO3, MgO) and covalent compounds (SiO2, GeO2, and B2O3), individually or as

mixtures, on perovskite formation in sol±gel derived lead magnesium niobate, PbMg0.33Nb0.67O3(PMN) were studied. The ionic additives

were coprecipitated with the PMN while the covalent compounds were added as ®ne particulates. At a level of 10 mol% additive per mol of PMN, additions of PbO, BaTiO3, PbTiO3, as alkoxide solutions, increased perovskite formation from 82% to 90, 91 and 87% respectively.

A combination of PbO (10±15 mol%) and BaTiO3(8±9 mol%) increased perovskite formation further to 98%. The addition of 9 mol%

MgO alone (also as an alkoxide solution) decreased perovskite formation to 77%, but a combination of PbO (10 mol%) and MgO (5 mol%) increased perovskite formation to 100%. The chemical effects of ionic compounds can be partially explained via the Goldschmidt tolerance factor. Compositions with high tolerance factors (>0.9) exhibited high (>90%) perovskite formation. Additions of 5 mol% GeO2, SiO2, and

H3BO3to modi®ed PMN (14 mol% PbO and 9 mol% BaTiO3), resulted in signi®cant decreases in perovskite formation, from 98% to 86,

80, 50%, respectively, when heated to 9808C for 8 h. The reduction in perovskite for undoped PMN, however, was nearly double (62% perovskite) when 5 mol% SiO2was added. # 2000 Elsevier Science S.A. All rights reserved.

Keywords: Lead magnesium niobate; Sol±gel; Additives; Perovskite; Tolerance factor

1. Introduction

Lead magnesium niobate, PbMg0.33Nb0.67O3 (PMN) is

a relaxor ferroelectric which possesses a high dielectric constant (20,000) and has been studied extensively for actuator and capacitor applications [1]. However, the for-mation of low dielectric constant (200) pyrochlore phase (Pb1.83Nb1.71Mg0.29O6.39) during the synthesis and loss of

PbO at high (up to 13008C) sintering temperatures has limited the full use of this material. Extensive research has been performed to prepare pyrochlore free PMN from solid state reactions of constituent oxides. Shrout and Halliyal [2] have reviewed the problems associated with pyrochlore formation. Swartz and Shrout [3] reported the elimination of the pyrochlore phase via the formation of a columbite structure of MgNb2O6®rst, then reacting it with PbO to form

PMN. Other researchers obtained perovskite PMN by using ultra pure starting materials [4], adding excess MgO (2± 10%) [5,6], excess PbO [7] or 4 to 5 mol% BaTiO3[8].

Low temperature ®ring (below 9508C) which would allow the use of less expensive silver palladium electrodes generally requires sintering aids which lead to lowering and aging of the

dielectric properties of PMN. However, lower sintering tem-peratures have been achieved for many sol±gel derived mate-rialsduetothe homogeneous mixingofmetaloxideprecursors (metal alkoxides) as solutions and the high activity of the resulting powders. Two research groups (Payne [9] and Roy [10]) have reported perovskite PMN formation at 800 and 7758C respectively from sol±gel method.

The objective of this study was to determine the effects of chemical additives on perovskite formation in sol±gel derived PMN. The selected additives were ionic compounds, e.g. PbO, BaTiO3, PbTiO3, and MgO, and covalent

com-pounds, e.g. B2O3 (added as H3BO3), SiO2 and GeO2,

individually or as mixtures. The ionic compounds are some which have been reported to increase perovskite formation or stability, while the covalent compounds are those which might be present in glasses used as sintering aids. The results are summarized and discussed in the following sections. 2. Experimental

2.1. Solution synthesis

PMN was synthesized according to the method of Roy group [10] with some modi®cations. To a 2 l three neck ¯ask *Corresponding author.

(2)

equipped with thermometer, stirrer and argon purge, 178.55 g (0.8 mol) of lead oxide (Hammond Lead Products, Inc.) was added and purged with argon overnight. Acetic acid (Fisher Scienti®c, 410.74 g, 6.8 mol) was added into the ¯ask (care-fully due to the resulting exothermic reaction). The mixture was heated to 1308C to distil out water and acetic acid (about 270 g total over a 4 h period). After the solution was cooled below 408C, 1025.07 g of methoxyethanol was added into the ¯ask, the solution was then again heated to 1308C to distil out methoxyethylacetate and methoxyethanol (about 300 g distillate) until the pH of the solution was 8.2 at room temperature. Magnesium ethoxide (Gelest Inc., 30.28 g, 0.27 mol) and niobium ethoxide (Cerac Chem. Co., 168.75 g, 0.53 mol) were added into the solution. There was a slight exotherm and the solution became cloudy. The solution was heated to 1308C until about 50 g of distillate (ethanol) was collected. Approximately 25% (weight of metal by weight of solution) clear solution was obtained.

2.2. Powder preparation

In a crystallization dish equipped with a magnetic stirrer, 10 g of deionized water was added to 100 g of the above PMN solution; gelation occurred quickly. The gel was dried at 1308C for 18 h and calcined at 3508C for 18 h to prepare the PMN precursor powder. X-ray diffraction showed a weakly crystalline pyrochlore structure after the 3508C calcine as shown in Fig. 1. The calcined powder was heated from room temperature to 1008C at 18C/min, from 100 to 5008C at 58C/min, held at 5008C for 1 h, then heated from 500 to 9808C at 58C/min, and held at 9808C for 8 h. 2.3. Additives on PMN

Additions of lead oxide, barium titanate, and magnesium oxide to PMN were performed in solution using lead acetate

in methoxyethanol, barium titanium alkoxide solution (Gel-est Inc.) and magnesium ethoxide in methoxy ethanol respectively. The additive solutions were mixed with the base PMN sol for 1 h prior to gelation. Gelation and powder preparation were carried out as described above for the PMN powder.

A modi®ed PMN powder containing lead oxide (14 mol%) and barium titanate (9 mol%) was used to study the chemical reactions with silica (precipitated from tetra-ethoxy silane), and boron oxide doped (23% by weight) silica (precipitated from tetraethoxysilane and boron tri-methoxide), germanium oxide (Gelest Inc.) and boric acid (Aldrich). The modi®ed PMN powder and 1±10 wt.% addi-tive were milled overnight in ethanol using a 60 ml poly-propylene bottle half ®lled (70 g) with high purity zirconia balls (5 mm diameter). The powder was then dried at 608C overnight, and heated from room temperature to 1008C at 18C/min, from 100 to 5008C at 58C/min, held at 5008C for 1 h, then heated from 500 to 9808C at 58C/min, and held at 9808C for 8 h.

2.4. X-ray diffraction study

A Philips automated powder diffraction system APD 3720 was used to examine the X-ray diffraction of heat treated powders. The relative amounts of perovskite and pyrochlore phases are determined using system software which compared relative intensities of known X-ray patterns for Pb(Mg0.33Nb0.67)O3and Pb1.83NB1.71Mg0.29O6.39.

2.5. Tolerance factor (t) calculation

The tolerance factor (t) of each composition was calcu-lated according to Goldschmidt [11] equation, t (RA

Ro)/(2)0.5_(R

B Ro), where RA, RBand Ro are the ionic radii

(3)

of cation A, cation B and anion oxygen respectively. The value of each ionic radius is listed in the Table 1.

3. Results and discussion

We have carried out the quantitative study of chemical effects on perovskite formation of sol±gel derived PMN. Goldschmidt [11] tolerance factor (t) of cubic perovskite structure was used to explain the chemical effects of ionic compounds PbO, MgO, BaTiO3, PbTiO3. The tolerance

factor is an indication of how tightly the atoms are packed in the cubic structure of ABO3 compound. The factor is

related to the ionic radii of atoms and is de®ned as the following:

t RA Ro=20:5RB Ro

where RA, RBand Ro are the ionic radii of large cation A,

small cation B, and of oxygen anion respectively. The most stable cubic perovskite structure of ABO3is formed when

its tolerance factor is equal to 1.000.

Barium titanate has a tolerance factor of 0.986 and exhibits stable 100% perovskite structure. For PMN, we assume Pb is a large cation and Mg and Nb are small cations. The ionic radius of Pb (1.20) is smaller then that of Ba (1.34) but the ionic radii of Mg (0.74) and Nb (0.69) are larger than that of Ti (0.61), thus the tolerance factor of PMN is reduced to 0.887 and exhibits only 82% perovskite. From the tolerance factor formula, the % perovskite of PMN can be increased by adding high tolerance factor compounds such as BaTiO3and PbTiO3or by increasing the amount of a

large radius cation such as lead ion.

Table 2 lists the calculated tolerance factor and the corresponding percentage of perovskite formation observed by X-ray diffraction for PMN with and without ionic additives. Fig. 2 is a graphical representation of these data. PbTiO3is not as effective as PbO or BaTiO3in increasing

perovskite formation. A 90% perovskite was obtained when 10% (by mol) or more of BaTiO3or PbO was added into the

Table 1

Ionic radii of selected ions

Ion Ionic radius (AÊ)

Ba2 _1.34a Pb2 _1.20a Oÿ2 _1.26b Nb5 _0.69a Mg2 _0.74a Ti4 _0.61a

a_{F.S. Galasso, Structure, Properties and Preparation of Perovskite Type}

Compounds, Pergamon Press, 1969, pp. 41±45

b_{F.A. Cotton and G. Wilkinson, Advanced Inorganic Chemistry, 5th}

ed., Wiley, 1988, pp. 1386±1388.

Table 2

Effect of ionic chemicals on the perovskite formation of sol±gel PMN (powder heated to 9808C for 8 h)

Sample Mol% of Chemical per mol of PMN Tolerance factor (t) % Perovskite

PbO MgO PT BT 1 0 0 0 0 0.887 82 2 10 0 0 0 0.930 90 3 16 0 0 0 0.956 92 4 0 9 0 0 0.858 77 5 0 0 3 0 0.889 86 6 0 0 12 0 0.892 87 7 0 0 0 3 0.890 88 8 0 0 0 8 0.895 90 9 0 0 0 16 0.901 94 10 10 5 0 0 0.924 100 11 15 5 0 0 0.945 97 12 8 0 0 8 0.927 96 13 8 0 14 8 0.927 98 14 14 0 0 9 0.951 98 15 13 0 0 13 0.948 98 16 14 0 0 17 0.953 98

Fig. 2. Observed perovskite content vs. Calculated tolerance factor for PMN modified with PbO, PbTiO3, MgO and BaTiO3(data from Table 2).

(4)

PMN. The amount of perovskite formation can be further increased to 98% by adding a mixture of PbO and BaTiO3.

The sol±gel PMN with excess MgO (9 mol%) exhibited low perovskite formation (77%), which can be explained on the basis of a low tolerance factor of 0.858. This does not agree with observations for PMN prepared from mixed oxides. Schulze et al. [5,7] showed that the fabrication of 100% perovskite PMN required excess MgO (at least 5 mol%). The observed percentage of perovskite differences between the sol±gel derived PMN and oxides derived PMN may be explained on the basis of a difference in the distribution of the excess MgO and its interaction in the formation of PMN. For the PMN prepared from oxides, the excess MgO was added as part of the columbite precursor MgNb2O6. The precursor helped to prevent the formation of

lead niobate pyrochlore from Pb and Nb. Schulze observed that after sintering, excess MgO existed as inclusions both as the grain boundary and within the perovskite grain. For sol± gel PMN, excess MgO was added homogeneously as an alkoxide solution. The excess is mixed on an atomic level in a material which has exhibited a weakly pyrochlore struc-ture after drying at 3508C (Fig. 1). Since the more stable pyrochlore phase is already present, the excess MgO is ineffective in preventing its formation. In addition, the incorporation of MgO in the perovskite structure decreases its tolerance factor over that of the stoichiometric PMN and one would expect the observed decrease in perovskite formation.

The combination of excess PbO and MgO in the sol±gel PMN generated a higher perovskite formation (97±100%) than the PMN containing excess PbO only (92% perovs-kite); this cannot be accounted for basis on the tolerance factor. The reaction kinetics are not considered in the calculation of Goldschmidt tolerance factor. The addition of MgO may react with niobium preferably to form the columbite precursor of MgNb2O6. This precursor prevents

the excess of PbO to react with niobium and to form lead niobate pyrochlore. Thus, the combination of excess of PbO

and MgO on the perovskite formation of PMN cannot simply be explained by the tolerance factor.

The combination of excess lead oxide and BaTiO3or a

combination of excess PbO, BaTiO3and PbTiO3in PMN

also exhibited high perovskite formation (98%) which can be attributed to their higher tolerance factors (>0.927). Although the tolerance factor cannot explain every case studied in this report (88% success rate), the factor can serve as an initial prediction tool before performing actual experi-mental works.

To study the potential impact of glass additives present in sintering aids such as those present in the sealing glass used by Srikanth and Subbarao [12], B, Si and Ge were added to the PMN precursor powder modi®ed with PbO (14 mol%) and BaTiO3 (9 mol%). As discussed earlier, the PMN

precursor powder is a poorly crystalline pyrochlore which, without additives, is not completely converted to perovskite upon heating to 9808C.

Srikant and Subbarao [12] used a sealing glass (Corning 7555) to lower the sintering temperature of oxides derived 0.93PMN-0.07PbTiO3(PMN-PT). They reduced the

sinter-ing temperature from 9508C/4 h to 7508C/30 min. The sealing glass contains 60±80% (by weight) PbO, 10±30% B2O3, 1±20% SiO2, 1±20% Al2O3and 1±20% ZnO. Despite

the high lead content of the glass, they found that pyrochlore formed at as low as 1% (by weight) glass at 7008C (20% pyrochlore after 4 h) and the amount of pyrochlore increased with time, temperature and glass content. Even at as low level as 3% (by weight) glass, 4 h at 8008C reduced the perovskite content to 0%. It was postulated that the destabilization of the perovskite phase could be attributed to the gradual depletion of MgO from the PMN-PT through the action of B2O3, SiO2 or Al2O3. They showed that, with

additions of MgO (MgO substituting 10 and 20% of the glass phase respectively), pyrochlore formation could be signi®cantly reduced or eliminated. However, ®red densities were reduced to less than 90% which may have changed the distribution of the glass or indicated a change in the glass

(5)

viscosity. Both factors could affect the reaction kinetics. Fig. 3 shows how the addition of covalent compounds such as B2O3 (as H3BO3), SiO2, B2O3 (23% by weight)

doped silica, and GeO2 reduced perovskite formation of

modi®ed PMN, which was 98% perovskite when heat treated without the additives. At 5 mol% concentration, addition of GeO2, SiO2and H3BO3signi®cantly decreased

the perovskite formation from 98% to 86, 80 and 50% respectively. For comparison, undoped PMN exhibited 62% perovskite when it was combined with 5 mol% SiO2. This level of additive roughly corresponds to the

combined amounts of B and Si in the Pb glass used by Srikanth and Subbarao. The sample produced 100% pyro-chlore in the 0.93PMN-0.07PT after 4 h at 8008C. The sol± gel powder began as a weakly pyrochlore phase (Fig. 1) and was heated nearly 2008C higher for an additional 4 h. The results indicated that a homogeneous addition of the PbO and BaTiO3in the sol±gel PMN produced a material much

more resistant to destabilization. The improved stability may be explained in part by the work of Hirata and Yamaguchil [13]. For interfacial reactions between BaTiO3

and Pb substituted BaTiO3with a PbB2O3glass, the rate of

reaction was governed by the solubility of the Pb and Ba in the glass. Because the Pb is more soluble than Ba, the Pb is reacted with glass instead of BaTiO3. In our work, in

addition to the increased tolerance factor, the addition of Ba to the PMN may reduce the rate of PMN reacting with the glass forming additives. Thus the weakly pyrochlore precursor powder was allowed to convert to perovskite before reacting with the covalent compounds. Also a Ba rich shell may develop, which is increasingly resistant to destabilization. As Hirata and Yamaguchi point out, how-ever, the interfacial reactions are also dependent upon glass volume, composition and distribution. Additional work would be required to determine the exact nature of the improved stability.

The B2O3doped silica did not have a greater impact that

the pure silica in destabilizing PMN perovskite even though B2O3alone had the greatest impact. At the 5 mol% level, it

had the equivalent perovskite content of 80% and continued to have a similar trend as the pure SiO2curve at higher

additive levels. Several explanations are available for this observation. First, the B2O3may have reacted with absolute

ethanol during milling to form volatile boron ethoxide which may have been lost during drying of the slurry or during the high temperature calcine. Thus, the remaining silica exhibited behaviour similar to the pure silica doped PMN. A second explanation is that boric acid melts at 1768C and may have become more homogeneously distributed around the PMN powder than would a higher

melting point B2O3 or borosilicate compounds, and as a

result the reaction kinetics were signi®cantly enhanced. 4. Conclusions

The chemical effects on the perovskite formation of sol± gel derived PMN has been studied. The addition of ionic compounds such as lead oxide, magnesium oxide, barium titanate and lead titanate or their mixture, to PMN increased the formation of perovskite. The results were explained by invoking the Goldschmidt tolerance factor of perovskite structure. Compositions with a high tolerance factor (>0.9) exhibit high (>90%) perovskite formation. The addition of covalent compounds such as boric acid, silica, boron oxide doped silica and germanium oxide, or their mixture, desta-bilized the PMN perovskite and reduced the formation of perovskite. Boron had the greatest impact on reducing perovskite formation followed by SiO2 and GeO2. The

B2O3 doped silica did not have an impact as as great as

pure silica in destabilizing PMN perovskite even though B2O3alone had the great impact. This may be the result of a

more uniform distribution of the lower melting point boric acid or possible loss of boron from the borosilicate due to the formation of volatile boron ethoxide. The PMN doped with PbO and BaTiO3showed a signi®cant improvement in

perovskite formation (80%) compared to undoped material (62%) for a 5 mol% addition of SiO2.

References

[1] I. Burn, Ceramic Capacitor Dielectrics, Engineered Materials Handbook, vol. 4, Ceramics and Glasses, ASM International, 1991, pp. 1112±1117.

[2] T.R. Shrout, A. Halliyal, Am. Ceram. Soc. Bull 66 (1987) 704. [3] S.L. Swartz, T.R. Shrout, Mater. Res. Bull 17 (1982) 1245. [4] J. Chen, A. Gorton, H.M. Chen, M. Harmer, J. Am. Ceram. Soc. 69

(1986) C303.

[5] S.L. Swartz, T.R. Shrout, W.A. Schulze, L.E. Cross, J. Am. Ceram. Soc. 67 (1984) 311.

[6] E. Goo, T. Yamamoto, K. Okazaki, J. Am. Ceram. Soc. 69 (1986) C188.

[7] H.C. Wang, W.A. Schulze, J. Am. Ceram. Soc. 73 (1990) 825. [8] D.H. Kang, K.H. Yoon, J. Mat. Sci. 26 (1991) 56±60.

[9] L.F. Francis, Y.J. Oh, D.A. Payne, J. Mat. Sci 25 (1990) 5007±5013. [10] P. Ravindranathan, S. Komarneni, A.S. Bhalla, R. Roy, J. Am.

Ceram. Soc. 74(12) (1991) 2996±2999.

[11] F.S. Galasso, Structure, Properties and Preparation of Perovskite Type Compounds, Pergamon Press, 1969.

[12] V. Srikanth, E.C. Subbarao, J. Mater. Res. 6(6) (1991) 1308. [13] A. Hirata, T. Yamaguchi, J. Am. Ceram. Soc. 80(1) (1997) 79±