Conduction of (Na,Ce) doped Bi 5 Ti 3 FeO 15 ceramics after different annealing processes

Abstract

Bi_5−xNa_x/2Ce_x/2Ti₃FeO₁₅ (BNTF-100x, x = 0, 0·05, 0·10, 0·20, 0·30, 0·40 and 0·60) ceramics were prepared via solid state sintering process. With x increasing, the resistivity increases at first and then decreases. The lower the concentration of oxygen vacancies, caused by more (Na,Ce) dopant adding, the less the deep trap energies in the band gap for the activated electrons to be mobile. For the samples containing high concentration of oxygen vacancies, the conductivities are higher when they are annealed in the air and oxygen than that in the vacuum. On the contrary, for the samples containing low concentration of oxygen vacancies, the conductivities of the samples in different annealing conditions behave quite the same way. Most (Na,Ce) doped Bi₅Ti₃FeO₁₅ ceramics exhibit intrinsic conduction at high temperatures with corresponding activation energies of 1·1–1·6 eV, while the extrinsic conduction has a close relationship with the concentration of oxygen vacancies and the activation ability of electrons.

Keywords

Bismuth layer-structured ferroelectrics Conduction Oxygen vacancy Annealing

Introduction

The bismuth layer-structured ferroelectrics (BLSFs), first discovered by Autivillius,1 are good potential candidates for non-volatile ferroelectric random access memory2, 3 and high temperature piezoelectric devices4 because they have high fatigue resistance and relatively high Curie temperature T_c respectively. The general formula of BLSFs is (Bi₂O₂)²⁺(A_m−1B_mO_3m−1)²⁻, in which A is a mono-, di- or trivalent element (or a combination of them) suited to dodecahedral coordination; B is a transition element with octahedral coordination, e.g. Fe³⁺, Ti⁴⁺, Nb⁵⁺, Ta⁵⁺, Mo⁶⁺ or W⁶⁺; and m is the integer, ordinarily varying from 1 to 5, indexing the number of the oxygen octahedral layers between every other neighbouring (Bi₂O₂)²⁺ layers.5, 6 The two vital problems of the application of BLSFs are their relatively low resistivity7, 8 and their limited piezoelectric activity.9 These problems can be improved by many methods, such as doping10, 11 or special processing technology.12, 13 The conduction mechanisms of BLSFs have been investigated by many researchers, but none have been widely accepted. Generally, the conduction of BLSFs can be divided into two parts: intrinsic conduction corresponding to the high temperature range and extrinsic conduction at low temperatures.

The effect of annealing atmospheres on the electrical properties of BLSFs has been extensively investigated.14^–17 However, the roles of different annealing atmospheres in the conduction process of BLSFs are still uncertain. For example, the leakage current of oxygen annealed Bi₄Ti₃O₁₂ single crystal showed higher values than that of an air annealed one at room temperature, as electron holes arising from the incorporation of oxygen molecules at oxygen vacancies acted as detrimental factors for resistivity.15 A similar phenomenon was reported on SrBi₄Ti₄O₁₅ thin films.16 Nevertheless, for the W⁶⁺ doped Bi₃TiNbO₉ ceramics, the roles of the two annealing atmospheres seemed to be opposite. The conductivity of the oxygen annealed Bi₃TiNbO₉ samples was lower than that of the as sintered ones.17

Bi₅Ti₃FeO₁₅ (BTF) ceramic is one of the BSLFs with Fe ion on the B site. Ordinarily, materials containing Fe ions have the problem of high leakage current due to the existence of Fe²⁺ and oxygen deficiency,18, 19 especially in the BiFeO₃ material. According to Scott et al.,20 oxygen vacancies had some relationship with the origin of space charges in some ferroelectrics. Qi et al.21 found that oxygen vacancies created deep trap energy in the band gap for the activated electrons to be mobile. Through the investigations on the Ti⁴⁺ and Ni²⁺ replacing of Fe³⁺, Qi et al.21 confirmed that oxygen vacancies rather than Fe²⁺ were the main reason of the high conductivity of the BiFeO₃ film.

In order to study the effect of Fe ions and oxygen vacancies on the conduction process in Fe contained BLSFs, the conductivities of (Na,Ce) doped BTF ceramics after different annealing processes were investigated. So far, there have been few detailed reports on the conduction of Fe contained BLSFs.

Experimental

Bi_5−xNa_x/2Ce_x/2Ti₃FeO₁₅ (BNTF-100x, x = 0, 0·05, 0·10, 0·20, 0·30, 0·40 and 0·60, hereinafter called BTF, BNTF-5, BNTF-10, BNTF-20, BNTF-30, BNTF-40 and BNTF-60 respectively) ceramics were prepared via conventional solid state reaction sintering. Bi₂O₃ (99·99%), TiO₂ (99·38%), Fe₂O₃ (99·00%), Na₂CO₃ (99·80%) and CeO₂ (99·99%) were weighted with the stoichiometric compositions and mixed by ball milling in ethanol for 8 h. The slurry was dried and calcined at 850°C for 2 h. The calcined powders were pressed into discs using a blinder, which was removed at 700°C for 1 h, and then they were embedded in the same component powders in a sealed Al₂O₃ crucible and sintered at 1060–1120°C for 2 h. The sintered samples were processed into 0·4 mm thickness and 8 mm diameter. To measure the electrical properties, electrodes were fabricated with a fired on silver paste. The BTF, BNTF-10, BNTF-20 and BNTF-40 samples were finally put into three different post-sintering annealing conditions: (1) sintered in air and cooled with the furnace, denoted as as sintered; (2) annealed in static oxygen at 700°C for 24 h (∼0·1 MPa) after sintering, denoted as oxygen annealed; and (3) annealed in vacuum at 700°C for 24 h (∼10² Pa) after sintering, denoted as vacuum annealed.

The dc resistance was measured over the temperature range from 150 to 550°C at a heating rate of 5°C min⁻¹ and a stabilising time of 30 min at a measuring temperature. An applying voltage of 50 V and a holding time of 1 min are adapted to measure the resistances of the order of 10¹⁴–10⁶ Ω using a high resistance meter (HP4329A; Hewlett-Packard, Palo Alto, CA, USA). A precision digital multimeter (Fluke 17B, Shanghai, China) was used to measure the relatively low resistances in the range between 10⁶ and 10 Ω.

Results and discussion

According to our earlier work,22 pure BNTF ceramics can be obtained, and their relative densities are all >94%. Most grains have a plate-like shape with a diameter of 2–5 μm and a thickness of 1–2 μm for BTF, BNTF-10 and BNTF-20 samples, and a diameter of 5–10 μm and a thickness of 1–2 μm for BNTF-40 and BNTF-60 samples.

Figure 1 gives the dependence of room temperature resistivity on the dopant content x in BNTF ceramics (x = 0, 0·05, 0·10, 0·20, 0·30, 0·40 and 0·60). With the dopant increasing, the resistivity of BNTF increases at first and then decreases. In other words, with the increase in grain size, and especially with the increase in diameter/thickness ration, the resistivity of BNTF ceramics substantially decreases. When x = 0·10, the resistivity reaches the highest value, ∼1×10¹³ Ω cm, which is about two orders higher than that of BTF.

Figure 1

Dependence of room temperature resistivity on the (Na,Ce) dopant content x in BNTF-100x ceramics (x = 0, 0·05, 0·10, 0·20, 0·30, 0·40 and 0·60)

Figure 2 shows the logarithms of the dc conductivities of BNTF-100x (x = 0, 0·10, 0·20 and 0·40) versus the reciprocal temperature and the linear fittings of BTF ceramics according to Arrhenius relationship. The conductivity of BTF is higher than that of the doped samples in the temperature range of 150–550°C, demonstrating that proper (Na,Ce) doping improves the resistivity of BTF. Two different lines of each sample can be obtained in Fig. 2, like the two fitting lines of BTF.

Figure 2

Logarithms of dc conductivities of BNTF-100x (x = 0, 0·10, 0·20 and 0·40) versus reciprocal temperature and linear fittings of BTF ceramics according to Arrhenius relationship

Figure 3 gives logarithms of the dc conductivities versus reciprocal temperature for BNTF-100x (x = 0, 0·10, 0·20 and 0·40) after different annealing processes. The conductivities of as sintered and oxygen annealed BNTF ceramics are quite the same, indicating that the conduction mechanisms of the two cases are similar. However, the conductivities of the vacuum annealed samples are lower. With (Na,Ce) dopant increasing, the conductivity differences between vacuum annealed samples and other samples become less obvious.

Figure 3

Logarithms of dc conductivities versus reciprocal temperature for a BTF, b BNTF-10, c BNTF-20 and d BNTF-40 after different annealing processes

From the ionisation energy point of view, the change of valence state of the Fe ions is much easier than that of the Ce ions. (The energy differences of Fe³⁺ to Fe²⁺ and of Ce⁴⁺ to Ce³⁺ are 1395·1 and 1598 kJ mol⁻¹ respectively.) Thus, the behaviours of charge carriers in BTF based ceramics are mainly the change of valence state of Fe ions, whether doping or not. On the other hand, the mobility of the electrons causes the change of the valence state of the Fe ions, so the conduction type of pure BTF ceramics is probable n type. As is known, the defect equation of (Na,Ce) substituting for Bi is shown in equation (1) (1) According to equation (1), when Na⁺ and Ce⁴⁺ substitute for Bi³⁺ in BTF ceramics, holes will be created under charge neutrality restrictions. These holes neutralise the electrons that appear in the material with n type conduction, such as the BTF ceramic, so that the concentration of carriers becomes smaller, which leads to lower conductivity. For BNTF-10, the hole neutralisation was complete. When the doping level is higher than that of the BNTF-10, the material became p type and more and more holes can be created and the conductivity becomes higher. The analysis above explains the variation of the room temperature resistivity with the change of dopant content, as shown in Fig. 1.

The temperature dependence of conductivity can be fitted by the Arrhenius equation (2) where σ₀ is the pre-exponential factor, E_a is the activation energy, k is the Boltzmann's constant and T is the temperature. Equation (2) shows that lg σ has a linear relationship with 1/T, and E_a can be calculated from the slope of the fitted line. The Arrhenius plots of as sintered BTF are given in Fig. 2, in which two regions with different values of E_a can be observed obviously. Other samples (the plots are not shown here) also behave at quite the same trend, and the corresponding values of E_a are given in Table 1.

Table 1

Activation energy E_a of BNTF ceramics at low temperature (LT) and high temperature (HT) for different annealing atmospheres

Samples	E_a-LT/eV	E_a-HT/eV
BTF as sintered	0·62±0·01	0·85±0·02
BTF oxygen	0·64±0·01	0·81±0·02
BTF vacuum	0·83±0·02	1·33±0·04
BNTF-10 as sintered	0·89±0·03	1·36±0·01
BNTF-10 oxygen	0·78±0·05	1·41±0·02
BNTF-10 vacuum	0·72±0·05	1·54±0·03
BNTF-20 as sintered	0·70±0·04	1·32±0·02
BNTF-20 oxygen	0·69±0·03	1·37±0·03
BNTF-20 vacuum	0·68±0·03	1·37±0·03
BNTF-40 as sintered	0·67±0·02	1·12±0·02
BNTF-40 oxygen	0·69±0·02	1·14±0·02
BNTF-40 vacuum	0·70±0·03	1·17±0·02

The generation or activation of charge carriers and the existence of possible paths for charge carriers’ mobility are the two main requirements to realise the conduction process. It is reported that Fe²⁺ would turn to Fe³⁺, and Fe³⁺ would turn to Fe⁴⁺ for the BiFeO₃ ceramic when it was annealed in O₂.23, 24 However, Fe³⁺ would hardly turn to Fe²⁺ when it was annealed in vacuum.23 This may shed a new light on the conduction of BTF based ceramics. The essence of the valence fluctuation of Fe ions is the activation of electrons. The valence fluctuation of Fe ions will be much easier in oxidising atmospheres as compared to that in the vacuum, which lacks the reducing atmospheres to make the Fe ion reduced. Therefore, the activation of electrons in oxygen annealed and as sintered samples will become easier than that in vacuum annealed samples. On the other hand, oxygen vacancy can be acted as the possible paths for charge carriers’ mobility. Our earlier investigations shows that (Na,Ce) doping can reduce the concentration of oxygen vacancies, especially for BNTF-20 and BNTF-40, which have much a lower concentration of oxygen vacancies than that of BTF and BNTF-10.22 Thus, there are less deep trap energies in the band gap for the activated electrons to be mobile in BNTF-20 and BNTF-40 ceramics.

From the above analysis, BTF and BNTF-10 can provide enough oxygen vacancies for the charge carriers to be mobile, while BNTF-20 and BNTF-40 provide limited oxygen vacancies. In the case of BTF and BNTF-10, because the concentration of oxygen vacancies is high and the activation of electrons in vacuum annealed samples becomes more difficult, the conductivities of the vacuum annealed samples are lower than those of other samples, as seen in Fig. 3a and b respectively. However, there will be much less deep trap energies in the band gap for BNTF-20 and BNTF-40. Thus, the conductivities of the samples will be almost the same after any annealing process, whether the electrons can be activated. This means that the conductivities of BNTF-20 and BNTF-40 in different annealing conditions behave quite the same way, seen in Fig. 3c and d respectively. Similarly, the fitting results of all the samples after different annealing processes by Arrhenius relationship are given in Table 1.

Table 1 shows activation energy E_a of different samples in different annealing atmospheres. For vacuum annealed BTF, E_a in the low and high temperature range are 0·83 and 1·33 eV, which are higher than those of as sintered and oxygen annealed BTF. This phenomenon is attributed to the high concentration of oxygen vacancies in BTF ceramics, and the activation of the electrons is prevented in the vacuum annealed samples. E_a in the low temperature range are all ∼0·70 eV for BNTF-20 and BNTF-40, while E_a in the high temperature range are ∼1·35 and 1·15 eV for BNTF-20 and BNTF-40 respectively, regardless of the annealing atmospheres. This phenomenon indicates that the behaviours of charge carriers at low temperatures are probably the same for the BNTF-20 and BNTF-40 samples. The conclusion that the conduction at high temperatures is much easier in BNTF-40 than that in BNTF-20 also can be drawn. E_a at high temperatures of BNTF ceramics are in the range from 1·1 to 1·6 eV, which are lower than those of other BLSFs reported. The E_a of intrinsic conduction is equal to the half the energy of the band gap.8 The band gap energies of Bi₄Ti₃O₁₂, BaBi₄Ti₄O₁₅ and BaTiO₃ are all ∼3·3 eV.25 The BTF ceramics can be considered as Bi₄Ti₃O₁₂–BiFeO₃ ceramics. The E_a of intrinsic conduction for BiFeO₃ based ceramics are in the range from 0·72 to 0·95 eV,26 so the E_a of intrinsic conduction for BTF based ceramics should be <1·6 eV. This is consistent with the values of E_a for BNTF ceramics at high temperatures shown in Table 1. In the low temperature range, the extrinsic conduction has a close relationship with the concentration of oxygen vacancies and the activation ability of electrons.

Conclusion

In summary, with the (Na,Ce) dopant content x increasing, the resistivities of Bi_5−xNa_x/2Ce_x/2Ti₃FeO₁₅ ceramics (x = 0, 0·05, 0·10, 0·20, 0·30, 0·40 and 0·60) increase at first and then decrease, indicating an n type conduction of BTF ceramics. When x = 0·10, the resistivity reaches the highest value, ∼1×10¹³ Ω cm, which is about two orders higher than that of BTF. When the concentration of oxygen vacancies becomes lower with more (Na,Ce) doping, there are less deep trap energies in the band gap for the activated electrons to be mobile. For the samples containing high concentration of oxygen vacancies, such as BTF and BNTF-10, the valence state of the Fe ions will change easily when the samples are annealed in the air and oxygen, while it will hardly change in the vacuum. Therefore, the conductivities of the vacuum annealed samples are much lower than those of the as sintered and oxygen annealed samples. On the contrary, for the samples containing a low concentration of oxygen vacancies, such as BNTF-20 and BNTF-40, the conductivities of the samples in different annealing conditions behave quite the same way. The activation energies of intrinsic conduction for BTF based ceramics are lower than those of other BLSFs reported, and the extrinsic conduction has a close relationship with the concentration of oxygen vacancies and the activation ability of electrons.

Footnotes

Acknowledgements

This work was sponsored by the International Cooperation Program of SAST (grant no. 09520709200) and the National High-tech R&D Program of China (863 program, grant no. 2009AA03Z108).

References

Autivillius

: ‘Mixed bismuth oxides with layer lattices. I. Structure type of CaBi₂B₂O₉’, Ark. Kemi, 1949, 1, 463–480.

Dearaujo

CAP

, Cuchiaro

, McMillan

, Scott

: ‘Fatigue-free ferroelectric capacitors with platinum-electrodes’, Nature, 1995, 374, (6523), 627–629.

Park

, Kang

, Bu

, Noh

, Lee

, Jo

: ‘Lanthanum-substituted bismuth titanate for use in non-volatile memories’, Nature, 1999, 401, (6754), 682–684.

Turner

, Fuierer

, Newnham

, Shrout

: ‘Materials for high-temperature acoustic and vibration sensors – a review’, Appl. Acoust., 1994, 41, (4), 299–324.

Subbarao

: ‘Family of ferroelectric bismuth compounds’, J. Phys. Chem. Solids, 1962, 23, 665–676.

Newnham

, Wolfe

, Dorrian

: ‘Structural basis of ferroelectricity in bismuth titanate family’, Mater. Res. Bull., 1971, 6, (10), 1029–1040.

Shulman

, Testorf

, Damjanovic

, Setter

: ‘Microstructure, electrical conductivity, and piezoelectric properties of bismuth titanate’, J. Am. Ceram. Soc., 1996, 79, (12), 3124–3128.

Zhou

, Dong

, Yan

, Chen

, Mao

: ‘Doping effects on the electrical conductivity of bismuth layered Bi₃TiNbO₉-based ceramics’, J. Appl. Phys., 2006, 100, (4), 044112.

Yan

, Zhang

, Ubic

, Reece

, Liu

, Shen

, Zhang

: ‘A lead-free high-Curie-point ferroelectric ceramic, CaBi₂Nb₂O₉’, Adv. Mater., 2005, 17, (10), 1261–1265.

10.

Yan

, Zhang

, Ubic

, Reece

: ‘B-site donor and acceptor doped Aurivillius phase Bi₃NbTiO₉ ceramics’, J. Eur. Ceram. Soc., 2006, 26, (13), 2785–2792.

11.

Yan

, Zhang

, Zhu

, He

, Yu

, Li

, Zhou

: ‘The effect of (Li,Ce) and (K,Ce) doping in Aurivillius phase material CaBi₄Ti₄O₁₅’, Mater. Res. Bull., 2004, 39, (9), 1237–1246.

12.

Tani

, Kimura

: ‘Reactive-templated grain growth processing for lead free piezoelectric ceramics’, Adv. Appl. Ceram., 2006, 105, (1), 55–63.

13.

Yan

, Ning

, Zhang

, Reece

: ‘Textured high Curie point piezoelectric ceramics prepared by spark plasma sintering’, Adv. Appl. Ceram., 2010, 109, (3), 139–142.

14.

Zhu

, Tang

, Ke

, Qiu

, Lei

, Hou

, Yan

, Xu

: ‘Annealing effect on relaxor behaviours of spark plasma sintered Pb(Sc_1/2Nb_1/2)O₃ superfine ceramics’, Adv. Appl. Ceram., 2011, 110, (2), 74–79.

15.

Noguchi

, Matsumoto

, Miyayama

: ‘Influence of oxygen vacancies on the polarization properties in Bi₄Ti₃O₁₂ ferroelectric single crystals’, in ‘Electroceramics in Japan IX’, (ed. K. Katayama et al.), 19–22; 2006, Stafa/Zurich, Trans Tech Publications Ltd.

16.

Simoes

, Ramirez

, Riccardi

, Longo

, Varela

: ‘Effect of oxidizing atmosphere on the electrical properties of SrBi₄Ti₄O₁₅thin films obtained by the polymeric precursor method’, Solid State Sci., 2008, 10, (12), 1951–1957.

17.

Zhou

, Li

, Yang

, Dong

, Yan

: ‘Effect of annealing on dielectric behavior and electrical conduction of W⁶⁺ doped Bi₃TiNbO₉ ceramics’, Appl. Phys. Lett., 2007, 90, (21), 212908.

18.

Wang

, Zhou

, Zhang

, Chen

, Liu

: ‘Room-temperature saturated ferroelectric polarization in BiFeO₃ ceramics synthesized by rapid liquid phase sintering’, Appl. Phys. Lett., 2004, 84, (10), 1731–1733.

19.

Palkar

, Pinto

: ‘BiFeO₃ thin films: novel effects’, Pramana – J. Phys., 2002, 58, (5–6), 1003–1008.

20.

Scott

, Araujo

, Melnick

, McMillan

, Zuleeg

: ‘Quantitative measurement of space-charge effects in lead zirconate–titanate memories’, J. Appl. Phys., 1991, 70, (1), 382–388.

21.

, Dho

, Tomov

, Blamire

, MacManus-Driscoll

: ‘Greatly reduced leakage current and conduction mechanism in aliovalent-ion-doped BiFeO₃’, Appl. Phys. Lett., 2005, 86, (6), 062903.

22.

Huang

, Wang

G.-S

, Li

Y.-C

, Liang

R.-H

, Cao

, Dong

X.-L

: ‘Electrical properties of (Na,Ce) doped Bi₅Ti₃FeO₁₅ ceramics’, Phys. Status Solidi A, 2011, 208A, (5), 1047–1051.

23.

Dai

, Akishige

: ‘Electrical properties of multiferroic BiFeO₃ ceramics synthesized by spark plasma sintering’, J. Phys. D: Appl. Phys., 2010, 43D, (44), 445403.

24.

Dho

, Qi

, Kim

, MacManus-Driscoll

, Blamire

: ‘Large electric polarization and exchange bias in multiferroic BiFeO₃’, Adv. Mater., 2006, 18, (11), 1445–1448.

25.

Ehara

, Muramatsu

, Shimazu

, Tanaka

, Tsukioka

, Mori

, Hattori

, Tamura

: ‘Dielectric-properties of Bi₄Ti₃O₁₂ below the Curie-temperature’, Jpn J. Appl. Phys., 1981, 20, (5), 877–881.

26.

Azough

, Freer

, Thrall

, Cernik

, Tuna

, Collison

: ‘Microstructure and properties of Co-, Ni-, Zn-, Nb- and W-modified multiferroic BiFeO₃ ceramics’, J. Eur. Ceram. Soc., 2010, 30, (3), 727–736.