Enhancement of tetragonal anisotropy and stabilisation of the tetragonal phase by Bi/Mn-double-doping in BaTiO3 ferroelectric ceramics

To stabilise ferroelectric-tetragonal phase of BaTiO3, the double-doping of Bi and Mn up to 0.5 mol% was studied. Upon increasing the Bi content in BaTiO3:Mn:Bi, the tetragonal crystal-lattice-constants a and c shrank and elongated, respectively, resulting in an enhancement of tetragonal anisotropy, and the temperature-range of the ferroelectric tetragonal phase expanded. X-ray absorption fine structure measurements confirmed that Bi and Mn were located at the A(Ba)-site and B(Ti)-site, respectively, and Bi was markedly displaced from the centrosymmetric position in the BiO12 cluster. This A-site substitution of Bi also caused fluctuations of B-site atoms. Magnetic susceptibility measurements revealed a change in the Mn valence from +4 to +3 upon addition of the same molar amount of Bi as Mn, probably resulting from a compensating behaviour of the Mn at Ti4+ sites for donor doping of Bi3+ into the Ba2+ site. Because addition of La3+ instead of Bi3+ showed neither the enhancement of the tetragonal anisotropy nor the stabilisation of the tetragonal phase, these phenomena in BaTiO3:Mn:Bi were not caused by the Jahn-Teller effect of Mn3+ in the MnO6 octahedron, but caused by the Bi-displacement, probably resulting from the effect of the 6 s lone-pair electrons in Bi3+.


A. Electron microscopy
Before this study, we observed secondary phases containing considerable amount of Mn at triple junctions of grains in relatively highly Mn-doped ( > ∼ 1 mol%) BaTiO 3 samples by scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The samples were prepared with raw powders mixed in an agate mortar by hand. Therefore, doping amount into BaTiO 3 was set to 0.5 mol% or less and a planetary ball mill was used for the raw powder mixing process in order to obtain homogenised samples. No secondary phase was observed in the samples by SEM and TEM.

B. Magnetic susceptibility
Since the Mn-contained secondary phases showed a magnetic order at low temperatures, temperature dependence of magnetic susceptibility for the highly Mn-doped BaTiO 3 samples deviated slightly from the Curie-Weiss law. In contrast, magnetic susceptibility of the samples in this study obeyed the Curie-Weiss law very well (R 2 > 0.9995) as indicated in Table 2. This suggests that formation of Mn-contained secondary phases is negligible.

C. X-ray powder diffraction
Even though the raw powders were well mixed using a planetary ball mill, diffraction peaks from undefined impurity phases were detected for the undoped and doped BaTiO 3 samples. However, no specific peak attributed to Mn-and/or Bi-doping was clearly found, and the intensity ratio of the largest unidentified peak to the largest peak from the main phase for the undoped BaTiO 3 (about 6:1000) was decreased to about 1:1000 by Mn-and/or Bi-doping. These are indicative of no segregated secondary phase directly accompanied by the Mn-and/or Bi-doping.

D. XAFS
As indicated in Fig. 3c and Table 1, Bi-L 3 EXAFS spectra were well fitted with the single coordination model of Bi at A-site in the perovskite-type BaTiO 3 . This suggests that Bi atoms existed almost in BaTiO 3 and those in secondary phases with other coordination should be rare.

E. Temperature dependence of permittivity
Although T C shifted by Mn-and/or Bi-doping, abruptness of the permittivity change at T C almost unchanged. This means that the dopants were distributed in the host BaTiO 3 homogeneously, because the permittivity change at T C would be gradual if concentration of the dopants were fluctuated.

F. Impedance (dielectric) spectroscopy
The dielectric permittivity data shown in Figs. 1a-d were derived from impedance measurements. The relation between the complex permittivity ε * = ε + jε (ε , ε : real and imaginary parts of permittivity, respectively) and the complex impedance Z * = Z + jZ (Z , Z : real and imaginary parts of impedance, respectively) are given by where f : frequency, A: sample area, and l: sample thickness. For electrical heterogeneous samples with serial connections of parallel resistance-capacitance (R-C) components (e.g., a serial connection of grain bulk and grain boundary), the total impedance is given by where R i , C i represents resistance and capacitance of ith component, respectively, and also imaginary part of impedance are written as In a frequency range of 10 2 -10 7 Hz, Z for undoped, Mn-doped, and Bi/Mn-doubly doped BaTiO 3 at room temperature obeyed well a relation of d log Z /d log f = −1, and also frequency dependence of ε for these samples showed almost constant. This means that these samples electrically consist of only one component with a large RC (2πf ) −1 , suggesting homogeneous capacitors (insulators). In contrast, impedance of Bi-doped BaTiO 3 showed relatively conductive behaviour consisting of two electrical components, bulk (b) and grain boundary (gb). Each resistance (R b , R gb ) and capacitance (C b , C gb ) were roughly estimated from Z * and the complex electric modulus M * = M + jM = 1/ε * of BaTiO 3 :Bi(0.4%) at room temperature: R b 10 4 Ω, R gb 10 1 Ω, C b C gb = 10 −9 − 10 −10 F. The huge dielectric loss of BaTiO 3 :Bi shown in Fig.  1d are owing to these row resistances, especially low R gb , which may not be caused by segregation of Bi-contained conductive layer at the grain boundaries, taking the XAFS result into account, but lowering of potential barrier height at the interfaces of Bi-doped bulk grains.
Considering the data and the discussions above mentioned, the doped BaTiO 3 samples in this study are homogeneous enough to be employed for the measurements and the analyses.

III. X-RAY POWDER DIFFRACTION AND RIETVELD ANALYSIS
As described in METHODS section, X-ray powder diffraction was carried out twice with a mixed powder specimen with the CeO 2 standard and a pure one for each sample. Diffraction patterns taken with the mixed and pure specimens of BaTiO 3 :Mn:Bi(0.2%) as a representative of the samples in this study are shown in Figs. S1 and S2, respectively. First, a diffraction pattern of the mixed specimen was refined by the two-phase Rietveld method with the fixed lattice constant of the CeO 2 standard, as shown in Fig. S1. Secondary, using the obtained lattice constants a and c from this refinement as fixed parameters, the single-phase Rietveld analysis for a diffraction pattern of the pure specimen was carried out, as shown in Fig. S2, in order to obtain c-axis atomic position z and isotropic atomic displacement factor B for each atom. Obtained crystallographic parameters for the samples in this study by the first and the second Rietveld analyses are listed in Tables SII and SIII, respectively. Some criteria of fit 1 for each refinement are also indicated in the tables.